Prijepis ovog videozapisa

Day 1 Session 1 Migration and fundamental rights

Shada Islam

Right. Thank you very much indeed for taking your seats.

Good afternoon everyone. Hello and welcome. Welcome to the 2023 European Network of Ombudsman Conference. A warm welcome to also all of you, many of you joining us online. I'm Shada Islam. I'm an independent EU analyst and commentator based here in Brussels. And it's my delight, privilege actually, to moderate this conference. So the hashtag is ENO Network and friends and colleagues, we have a very exciting, a very informative afternoon ahead of us.

We are going to be focusing on the two things that are so essential, so pivotal, so critical to us, human rights and citizens, right? We're going to be talking about migration and artificial intelligence, how it impacts on us.

We are privileged because we have with us Sally Hayden. She is the author of a fantastic book that all of you should be reading if you're looking at human rights and citizens' rights and our migration policy. The book is called My Fourth Time We Drowned, seeking refuge on the world's deadliest migration route. Sally is the Africa correspondent for the Irish Times and her lecture will be followed by a panel discussion.

Then there will be a coffee break and we'll return for the second panel, which is on lessons from public administrations using artificial intelligence. This will be opened by Michael O'Flaherty, director of the EU's Agency for Fundamental Rights. Welcome, Michael.

But before all of that, I'd like to give the floor to Emily O'Reilly, European Ombudsman, for her opening address to this conference. Emily, the podium is yours.

Emily O'Reilly

Thank you very much, Shada, and good afternoon, everybody, and a very warm welcome to the European Network of Ombudsmen Conference and thank you for taking the time to join us today.

Now, it is a truism that the European Union is always experiencing some form of crisis. And the last decade and a half has certainly been witness to that, with the financial crisis, the migration crisis, the decision of the United Kingdom to leave the EU, the Covid pandemic, the war in Ukraine, and now the Israel-Gaza conflagration and its humanitarian and global implications.

We also watch with some trepidation the growing reach of artificial intelligence, while the literal existential crisis of climate change is increasingly impacting on our lives. These crises often represent a failure of politics, but also of imagination and of empathy. The pursuit of power over others, whether political, economic, territorial, or technological, lies, as it has always lain, at the heart of global upheaval and of threat. As Ombudsman, our capacity to act may at times appear to be limited, marginal. But our work takes place at the centre of power structures, at the heart of public administrations, and it is always our duty to illuminate the way that those power structures operate, and constantly to call them out when their actions are harmful to the people that they serve.

We hold a mirror to those actions so the world might better observe the forces that control so much of our lives. That power of illumination, of forcing a clear-eyed look and evaluation of the reality and impact of political and administrative decision-making on the lives of ordinary people, is what underpins the work of our first speaker: Sally Hayden, an Irish and international journalist, and the author of a quite remarkable and multi-award winning book, My fourth time, we drowned.

For those of you who have not read it, I would recommend you to do so. It is a book that cuts through the abstractions of migration and asylum-seeking through the simple but powerful act of talking directly to the people attempting to make their way to Europe. It illuminates specifically the often cruel and brutal way those people are treated in camps and at crossings, and particularly in Libya. Most critically, it forces all of us in this Union to examine the extent to which decisions that we make play a role in that misery, in that brutality. And it does that by not allowing us to turn our faces away from the human, from an appreciation of our common humanity that is supposed to animate the direction and the purpose of the European Union.

Sally Hayden does not provide answers to these problems. That is not, she says, her role. Rather, the journalist provides the raw material, the raw facts, for those whose purpose it is to provide those answers if they can. And I think we can all agree that those answers are not simple.

For as long as there is war, for as long as there is hunger, for as long as there is oppression, for as long as the climate continues to wreak havoc on lives and livelihoods, people will choose to migrate to places of greater safety and greater opportunity. Just as they have done since the dawn of time, and just as you and I might also choose to do. And sometimes they will migrate, as expressed by one young man in Tunisia a few weeks ago, as he gazed across the Mediterranean, simply because “you in Europe have so much and we have nothing”.

As European Ombudsman, my office is currently investigating two particular aspects of this issue, both of them in the context of EU accountability.

The first concerns the actions of Frontex, the EU border agency, in the lead-up to and the aftermath of the capsizing of the Adriana fishing vessel in June of this year in Greek waters, where more than 500 people drowned, the majority of their bodies never yet recovered. We are specifically looking at the responsibilities of Frontex, both legal and ethical, when it comes to the operation of its search and rescue missions. Greek Ombudsman, my colleague Andreas Pottakis, is also investigating the role of the Greek authorities and particularly its Coast Guard, and we shall hear from him later.

Separately we are looking at the EU-Tunisia deal, signed earlier this year by Commission President von der Leyen and Tunisian President Saied, in the company of the Dutch and Italian Prime Ministers, essentially a commitment of EU funding in exchange for Tunisian cooperation in the prevention of attempts by migrants to cross to the EU from Tunisia.

We are specifically looking at what the EU has committed to, or will commit to, vis-à-vis the fundamental rights protection of those people. In all of this, we see the struggle between the upholding of the foundational values of the EU and what happens to those values when faced with the raw and often brutal politics that command so much of the narrative around migration, asylum and integration. Is it a neither nor? If we place the high value that we should on protecting fundamental rights, do we allow a space for bad actors to enter and to weaponise migration with even more devastating consequences? How do we navigate to protect our European soul? The ethical challenges are obvious. The politics are complex and at times intractable.

Our panel will later probe all of these issues in more detail, but first it is my great pleasure to invite Sally Hayden to the podium. Thank you.

Sally Hayden

34 years ago, on this exact days, the Berlin Wall fell. 1989 was also the year I was born. It was a time, I believe, when there was huge optimism about the future of Europe and what the EU itself could stand for. Today, what I see is a reversal of that hope. I see the condoning of gross human rights abuses, hypocrisy from European leadership and the normalisation of mass death on our borders. I see EU policies that, outside the EU, are propping up dictatorships, militias and other systems that oppress people further and, in the long term, are increasing the reasons why they need to flee.

I am a journalist, not an activist, and I don't propose specific changes. But as a journalist, my job is to speak truth to power and to point out gross wrongdoing. Aside from being a journalist, I am also a European... And over almost a decade now of reporting on migration-related issues, I have become increasingly horrified and ashamed about what is being done in my name; at the structures that are being built up to silence those undergoing abuse and the rhetoric that is being used to distance our citizens from the human consequences of what we are responsible for.

During this speech, I am going to talk about some of what I have found out, but there is obviously limited time and I hope attendees will consider reading my book, which explains everything in more detail.

First, to tell you a bit about me and why I am talking about this. I am a correspondent for the Irish Times, but I have also reported from across Africa and the Middle East for a wide range of other media, including The Guardian, CNN, Al Jazeera, Channel 4 News, The Washington Post and The New York Times.

In 2018, I was unexpectedly contacted by refugees incarcerated in Libyan migrant detention centres who were using hidden phones to appeal for help. Since that day, I have focused on what is happening to people along what is known as the Central Mediterranean Migration Route between Libya or Tunisia and Italy or Malta. The UN has called it “the deadliest migration route in the world”. Since 2014, more than 28,200 men, women and children have died or gone missing on the Mediterranean Sea while trying to reach Europe, more than 22,400 of them along this route. It is a staggering number that is also likely to be an underestimate.

The main reason I focused on the Central Mediterranean is because of the EU's role. In 2017, the EU and Italy began what appeared to be a deliberate circumnavigation of International law aimed at keeping people off European territory. We carry out surveillance, flying drones and planes to spot boats while spending tens of millions of euros on supporting the Libyan Coast Guard to intercept them. International law says a European vessel cannot return people to Libya, so we pay for the Libyans to do it instead.

In less than seven years, more than 125,000 people have been forced back to a militia-run state where they are often locked up indefinitely in detention centres that Pope Francis, among others, have called concentration camps. They are not charged with anything, not given access to a lawyer or a way to appeal their imprisonment. Among them are many children who go without an education. Women, often pregnant as a result of rape, have given birth in the detention centres without medical care.

Many fled wars in countries like Somalia or Sudan, or dictatorships like Eritrea. In countries neighbouring theirs, they were extorted or abused by security forces, threatened with deportation or made to live in refugee camps without enough to eat. When they try to travel further, they are held for ransom by smugglers and traffickers, their families forced to sell everything they own to raise thousands of dollars. International human rights lawyers say the EU-supported interceptions at sea and return to detention in Libya could be considered crimes against humanity.

My book was, among other evidence, cited in an international criminal court submission by the ECCHR last year that named 16 high-level decision-makers from EU member states, the EU Commission, Frontex, the European External Action Service and the EU Naval Force Operation as alleged co-perpetrators.

I began documenting what was happening to people caught at sea, because I realised European officials seemed to be deliberately not doing it. One question I asked multiple times, including to officials here in the European Parliament, was how many people die in detention. Frontex told me it is not in their mandate to track this, as Libya is a sovereign state. EEAS told me to ask the UN Refugee Agency. UNHCR had no comprehensive figures.

I myself documented large numbers of deaths from starvation, medical neglect, shootings, bombings and torture. The dead included at least 53 people killed during a bombing and buried in unmarked graves. They included detainees who were praying when a militia opened fire on them. They included a young boy who died of treatable appendicitis and his father who died weeks later of a heart condition and apparent heartbreak. There are so many more.

These deaths are the human consequences of our anti-migration efforts, but where is the accountability? Returning people to Libya traps them in a cycle that also involves human smugglers. The smugglers work in league with both the Coast Guard and the detention centre management. This has been documented by an independent UN fact-finding mission, as well as by me. Videos of captives being tortured are even circulated by their families on social media in a desperate bid to raise ransom money through crowdfunding.

Eritrean journalist Meron Estefanos said that around 1 billion euros in ransoms could have been paid to smugglers in Libya by now. By funnelling people back to Libya we support this monetisation of detention and we increase the reasons why Libyans need to flee themselves.

While reporting this book, I spent Christmas on a rescue boat in the Mediterranean. The people it rescued in the end were Libyans escaping the militias, the same groups that EU money supports. On that trip, I also witnessed first-hand the criminalisation of search and rescue today amid the absence of European-led search and rescue efforts. I myself spent a year under criminal investigation after the mission. What I am telling you about Libya is just one example. Abuses are happening along the length of our borders, from Greece to Melilla.

The debate around migration often makes it sound like the only fault of the EU is inaction, such as failing to carry out search and rescue missions. But there has been prolonged and decisive action.

Europe has exploited the wealth and resources of other countries for centuries, for example, and that has contemporary consequences, which need to be acknowledged. Almost everyone I speak to who is trying to cross the Mediterranean Sea comes from a former European or British colony.

I was asked today to speak specifically about why people are trying to come here, but there is no single answer to that.

The so-called migration crisis needs to be reframed as a global inequality crisis. We need to understand that a large portion of the world's population cannot get on planes, they cannot access visas, those trying to move are not a homogenous group. They come from a huge range of different backgrounds and have many different reasons for needing to travel, including escaping conflict, persecution, corruption, the breakdown of institutions and crushing poverty, or being reunited with family members.

Among them are musicians, engineers, dentists, farmers, Law graduates, former aid workers, and people with a vast range of other experiences and skills. The only thing they all have in common is that they do not have the security offered by a powerful passport, which they will get if they become a European. Though they do not garner adequate global news coverage, there are recent or ongoing conflicts everywhere from the Democratic Republic of the Congo, where nearly 7 million are displaced, to Tigray in northern Ethiopia, where estimates say hundreds of thousands could have been killed. There are people escaping Islamist insurgencies in the Lake Chad region, or insecurity in South Sudan. There are LGBT people looking for a place where they can live freely.

It is not that everyone is coming here, but those who do are seeking safety and security, and the chance to support people they have left behind.

Excluding Palestinians, 75% of the world's refugees are hosted in low and middle-income countries. Before the Ukraine war began, less than 10% of the world's refugees lived in the EU, while more than 26% are currently hosted in Africa. Others, who would likely not qualify as refugees, may still be suffering in ways exacerbated by Europe.

In West Africa, for example, we see overfishing by European vessels contributing to the destruction of local fishing industries, a key reason why many Senegalese fishers tell me they can no longer make a life at home.

Climate change is also expected to cause huge future displacement. In Somalia, during two visits last year, I saw first-hand the result of a drought that is said to have played a role in 43,000 excess deaths in 2022, half of them children under the age of five. The drought is said to have been climate change-related, yet Somalia itself produces a tiny fraction of the emissions that Western countries do. Despite having one-third of the population, my country, Ireland, produces around 51 times Somalia's emissions, according to the World Bank. Germany, with about five times the population, produces more than 900 times more.

At the moment, when you don't have a powerful passport, legal travel routes as they exist are racist and discriminatory. This is true even for the elite of many African countries, who have no plan to stay here. In September, I was at the Global Investigative Journalism Conference in Sweden. Some African journalists could not attend because they were denied visas, while others could barely participate because of visa restrictions. These are common and constant constraints for Africans in all industries.

For their part, European politicians talk about tackling the business model of smugglers, but this argument does not make sense.

Someone once described migration routes to me as being like balloons. When one is squeezed, another will expand.

Since the central Mediterranean became harder to cross, the so-called Atlantic sea route to Europe has gained in popularity. Nearly 7,000 people have died or gone missing attempting it since 2021, according to Spanish NGO Walking Borders. Their boats can end up floating at sea for months, as everyone on board dies one by one of thirst and hunger. Some, filled with corpses, wash ashore in the Caribbean and Brazil.

Smugglers are fulfilling a need. Without safe and legal routes available, people trying to move say they have no other choice. The more difficult our borders become to cross, the more the smuggling industry is fuelled.

Meanwhile, and this point is very important, EU money is arguably destabilising the region. The EU Trust Fund for Africa has seen 5 billion euro pledged to it since 2016, effectively to try and stop migration from 26 African countries.

It has been widely criticised for its lack of transparency and how much is being spent on securitisation, oppressing people further. My book details the funding and empowering militias in Libya.

Another example took place in neighbouring Sudan, where EU anti-migration money is said to have emboldened the Rapid Support Forces, the paramilitary group which has now gone to war, resulting in the displacement of nearly 6 million.

Many of those I met during a reporting trip to Tunisia in August were Darfuris, who said they were escaping a new genocide perpetrated by the RSF and are now trying to reach Europe too.

In Tunisia, President Kais Saied, accused of growing authoritarianism, is benefiting from European efforts to court him and potentially large sums of EU money. Black Africans were evicted from their homes and fired from their jobs after comments he made in February. Since then, some have been rounded up and dumped in the desert where they were left to die. Could Tunisia become the new Libya? We are waiting to see.

I talk all the time to EU staff who anonymously say they are desperately unhappy with the current situation on Europe's borders. They say stopping migration is being prioritised above all else and without proper regard for human rights principles. Right now, I am speaking for them too.

The annual budget for Frontex has increased to 845 million euros from 142 million in 2015. Frontex director Fabrice Leggeri stepped down last year over concerns the agency is not respecting human rights. But that does not seem to have prompted a real reckoning regarding how Frontex operates.

I know this is something the Ombudsman has been actively challenging.

UN agencies which receive EU funding also play a role. While EU officials call the UN's presence a safeguard, I have talked to a lot of unhappy UN staff who say they are being used to whitewash a system of brutal human rights abuses in Libya, for example. They say they cannot speak honestly about the effects of EU policy, or their own ineffectiveness and lack of oversight because their leadership want to keep receiving EU funding.

Anonymously, UN staff worry about their own lack of accountability. Refugees are also frightened to report bad experiences with the UN because they feel so vulnerable to retaliation. In general, and this is probably the most important thing I will say in this whole speech, I am always surprised by how few people with direct experience of forced or irregular migration are being included in conversations about it at all levels.

This disconnect, I believe, is incredibly problematic. Their voices need to be listened to, both their experiences and their policy recommendations. That is the only way to understand what is really happening, and by opening and maintaining lines of communication, you might be surprised by what you find out.

Speaking to refugees was how I first heard about allegations of corruption within the UNHCR refugee resettlement scheme in Sudan, for example, with staff accused of taking bribes of tens of thousands of dollars to facilitate legal resettlement. Following my investigation and an internal probe, one staff member was found guilty of soliciting bribes and abusing power, though refugees complained that corruption continued.

That is just one case underlining how little faith in the existing systems there is among refugees, something you must understand if you are involved in migration policy today. As a former UN investigator told me, when there is huge demand and small supply, there is always scope for exploitation.

And as we all know, the numbers of refugee resettlement spaces offered by Western countries are woefully inadequate. According to UNHCR, less than 40,000 refugees were resettled to safe countries in 2021, for example, though there were more than 27 million refugees worldwide.

One question I keep asking myself is how has the scale of abuses and mass debt on Europe's borders been allowed to happen? Dehumanisation is taking place and we are all responsible.

We urgently need to start questioning our use of language. Refugee is a legal status and migrant is a descriptor, but these are people like everyone else, with hopes, dreams and families who love them.

When you hear the term migration management, do you equate it with torture or indefinite detention? What about the term economic migrant? For many people, poverty is a threat to life. It can mean starvation, as I saw in northern Uganda during the early COVID lockdowns. It might mean watching family members die for other preventable reasons all the time. Today, life expectancy for people in sub-Saharan Africa is around 20 years less than it is for those in Europe.

The word migrant is being used to target and dehumanise a particular group of people, those without strong passports, who almost always have black or brown skin. Otherwise, why would we immediately launch a rescue for a sinking ship filled with tourists, but not one with migrants on board? That is where the hypocrisy comes in.

In June 2020, the European Parliament passed a resolution saying Black Lives Matter. But what about the mass deaths of black people on Europe's borders? Without migration, projections say that the EU's population will fall from 453 million in 2026 to 320 million by the end of the century.

Economists and other analysts say the EU needs migration. So why torture those who are enthusiastic to come and make a life here? Why inflict trauma they will live with forever if they are lucky enough to survive the journey in the first place? Over previous years, when I spoke to Africans who wanted to reach Europe, they had the idea of Europe as a place where human rights are respected, but that image is being shattered.

And abusive policies are contagious. They are being picked up and expanded on across the West and used by dictators and autocrats in other regions as justification too.

The speaker after me was originally supposed to be the European Commission Vice President for Promoting Our European Way of Life. In 2019, when Ursula van der Leyen was asked to justify this job title, she said one of our founding principles is that everybody has the same rights. But my reporting shows this is not true.

It seems to me that we are at a pivotal point in European history, and we need to ask more existential questions.

Like what kind of Europe do we want to live in? Is it one, as an August New York Times front page said, where mass drowning has been normalized? Do we still believe in human rights for everyone? And what are the long-term ramifications of the harms already done? We need to ask how future generations will judge us. Are we the people who reacted to a global inequality crisis by brutalizing the most vulnerable and condoning their murder? Some people tell me they have lost hope that anything will change, but hope should not be a precondition to doing the right thing.

You can stand on the right side of history, even if you are hopeless.

So to summarize, there are many people travelling for different reasons. They are fleeing wars, dictatorships, crushing poverty, climate change, corruption, and trying to be reunited with family members. They often come from countries that have long been exploited by Europe.

What the people I speak to want is a safe and dignified existence. Instead, they are facing death, torture, and other kinds of abuse on our borders in ways that have implicated European officials and the European public in alleged crimes against humanity. And now compare the 1.3 million people who claimed asylum in the EU in 2015, the year of the so-called migrant crisis, with the more than 4 million Ukrainians we have accepted with open arms. The reception for Ukrainian refugees showed us that a more empathetic policy has been possible all along, but so far that has not been extended to people from other regions. Instead, 34 years after the fall of the Berlin Wall, we are erecting more barriers than ever.

Thank you for listening. Thank you.

Shada Islam

Thank you very much indeed, Sally, for that very, very powerful presentation, bringing home to us here in the European Parliament the grim reality of what is happening on our shores... And you have humanised the voiceless, given a voice to the voiceless. Thank you very much indeed for that.

Now we will start the panel discussion to discuss many of the points on accountability, corruption, and others that Sally has raised.

Let me ask the European Ombudsman, Emily O'Reilly, to come up for the panel. Also, Ms. Tineke Strik, Madam Parliamentarian, Member of the Parliament's Civil Liberties, Justice, and Home Affairs Committee, please come to the podium as well. Monique Pariat, Director General at the European Commission's DG for Migration and Home Affairs, please come here as well. Andreas Pottakis, the Greek Ombudsman, looking forward to seeing you here as well. And last but not least, Reinier van Zutphen, the Dutch National Ombudsman.

Thank you very much indeed for joining me here. So before we begin, and you take your (head)phones... Hello. Nice to see you again.

So the panellists, please only turn on your microphones when you are going to speak and then turn them back off afterwards.

Monique, take the, exactly, take your, how are you, take your headphones if you need them. Right, so let's start. As I said, a very, very powerful presentation by Sally. She has talked about the numbers, of course, over 28,000 migrants and refugees who die on our shores trying to get in, not being able to, and then sent to detention camps.

She has talked about the lack of accountability. She has talked about the monetization of detention, the criminalization of those search and rescue missions, and she has called it a global inequality crisis. And a very important point, Sally, that you have made is that refugees and migrants are often, very, very often, not included in our conversations.

So I am going to start with you, Monique, and ask all of you, actually, to give your first reactions to what you have heard from Sally. And, of course, Monique, what is your response to what you've heard? Thank you very much.

Monique Pariat

It is very difficult for me to reply after Sally's because I am really the incarnation of all the evil that Sally has described. So, and since I am representing Vice President Schinas, who was unfortunately not able to come, so I would like, well, honestly, I think what Sally is describing is reality, and I will not deny that. I am sure, and I have been dealing with humanitarian assistance before, and I have been in many camps, and I know what the situation is there.

So it is the situation. And when Sally mentions global inequality crisis, this is clear. We are living in a rich Europe, and a large part of the world is not living under the same conditions.

So they are living in countries where they have, many people have no possibility to make a living, where people are living in failed states with corrupt governments, where there is no way for them to see a future for their children. So the temptation to move forward is obvious, and this is the reality.

That said, I would like also to mention we have more or less every year, well, last year at least, about 300,000 people that arrive irregularly. We need to add to that indeed all the peoples that have tried to come and that were prevented for one reason or the other, unfortunately, sometimes because they drowned in the Mediterranean.

But next to that, we have also in average 3 million people that every year migrate legally to the EU, and that obviously are not creating problems, that are welcome because they fulfil indeed also needs, and for the European economy, they are integrated, and they migrate totally legally.

We have also welcome, as Sally mentioned, 4 million Ukrainians, so that also has to be reminded.

The European Union has very strong fundamental rights protection.

This, I think the set-up of it, starting with the fact that all EU member states are part of the Geneva Conventions, that we are part of the Convention of Human Rights, that the Charter of Fundamental Rights is part of the EU Treaty, all this applies to all EU policies.

If this is applied correctly everywhere, obviously not, but we are looking at that. It is also your role to help us to do it, and there are many questions and points that were raised that I am very happy to mention, but maybe I would like to stop here as introductory message.

Shada Islam

Right, thank you very much indeed Monique, and we will come back to some of the issues that Sally has raised and other panellists will raise.

Let me just say that after our initial conversation that I will lead with questions, we will have time for questions and very short questions and very short comments from the floor.

Let me turn to you now, Tineke. So we have seen and heard from Sally, you yourself have been working on these issues and we have talked about it ourselves, the two of us before. We have also just seen a deal whose legality is circumspect, to put it mildly, between Italy and Albania. We have seen the Germans very recently coming out, I think just a couple of days ago, saying something similar will be done. Of course, Britain is trying to do something with Rwanda, we know that, on refugees.

So from your point of view, what you have heard from Sally and your own experience, what would be your comments? Thank you.

Tineke Strik

Thank you very much, first of all, for organizing this wonderful panel and also for this wonderful speech of Mrs. Hayden, who was really on spot, I would say. I really admire how she managed to really show the whole threat of all the things that are happening at the moment and how they relate internally.

Because it is really true, there is so little attention for the impact on individuals, on individual lives, of the current policies. And if you look at, if you zoom in on the impact, it really contradicts, you know, the values on which EU policy should be based upon. And it is not that we don't know the facts, that's the most problematic of it.

Because also, Sally, she really points at her own book, but also like the UN fact-finding mission report I was really thinking when that report came out in March, I thought now something will happen, because no one can deny any more how devastating the impact is of our funding to Libya. And even more how it contributes, our own funding, to the crimes against humanity. But it stayed silent afterwards.

And this is, I think... I mean, the Ombudsman, one of the things is accountability. This is really the key gap, I would say, the lack of accountability in all those violations of International law, but also EU law.

What we are fighting a lot on, also against the Commission actually, is the lack of enforcement of the EU rules that we have. We are not living in a vacuum, in a legal vacuum, but we really live in a vacuum of enforcement. And I think this atmosphere of impunity has gotten us in the situation where we are in now.

But talking about this externalization, actually, that you asked me, this deal with Albania, which also fits indeed a little bit in the deal with Tunisia and the funding of Libya, with different implications, of course. But it's one of the phenomenon of externalization.

And I would like to remind you on the report that we adopted in the Parliament two and a half years ago. It was on the human rights aspects of external dimension of asylum policies, where we made clear that you should always have an impact assessment on fundamental rights before you enter into cooperation, and make sure there is a transparent and independent monitoring of the impacts of such a cooperation.

Now, if I look at this MOU (Memorandum of Understanding), I see it's... I did not make a legal assessment yet, you know, thoroughly, but what I immediately think of, how can it be in compliance with our own EU law? With the secondary law, like the obligation for asylum seekers that we are responsible for to host them on our territories, to let them remain on our territories? And to make sure that, for instance, the reception conditions also on detention apply, so that not everyone can be automatically detained, but also the Charter of course, which makes sure that if you have legal control, if you have a legal responsibility, that you need to be able to comply with the EU standards and obligations.

So I don't see how this can happen, how this can be made true in the territory of a third state, which is not subject to EU standards. And also if I think of, if you process people who are in detention in another country, you know... How are you going to ensure proper procedures with sufficient judges, with sufficient legal aid provisions, and is that then according to the EU standards or not? So it leads to a lot of complicated questions, which we don't have an answer yet, so we also ask the Commission to take a stance on this, and it comes at a specific moment that the pact, the migration pact, is now being negotiated, which is also very odd, that you allow a Member state to, you know, to violate the current EU rules, which are also not foreseen in the pact.

So, yeah, this actually underscores the impunity that we are facing now.

Shada Islam

I think, to be fair, the Commission hasn't given its legal response yet to the pact, so I'll come to you in a second, in the second round, I think that is still being looked at. But you've made some very, very important points about will the conditions on foreign territory meet our standards.

Emily, so you introduced Sally, but I'd like to get some of your reactions to what she said, and especially the point about your own work that is being done with Frontex and also the accountability issue that's been raised by Sally.

Emily O´Reilly

Thank you, Shada.

I'd known about Sally's book for quite some time, because I'd read about it, I'd read about the awards that it got, but I hadn't read it, and I'd almost avoided reading it, because I knew it was going to be difficult to read. But then one of my colleagues read it and recommended that I read it, and when I read it, I was immensely impressed and moved by it.

I am a former journalist, but I don't think I would have ever had the courage to do what Sally has done. And she's about the same age as my eldest daughter, and so when I met her yesterday for the first time, I was asking what safety precautions did you take, you know, what did your family think? And so I could really appreciate what an incredible piece of work it is.

That doesn't mean that you have to agree with everything that's in it. At all.

But I think what is most powerful about the book is that it's not done from a distance. She spoke to people, she spoke to real people, she recounted real experiences. How you then analyse and rationalise and all of that? Well, you know, everybody can do that, but I just want to make that point about Sally's book.

But I also think that, you know, the Commission or the Council or Parliament don't act in a vacuum, and therefore I don't think, as Monique said, she feels or whatever (that they are), the incarnation of evil, the Commission is doing an incredibly difficult job with challenges popping up everywhere and trying to do as they will. But they don't work in a vacuum, you know.

And I think Sally, while she may have been pointing the finger at certain institutions and so on, also talked about we, about us, about us as Europeans, who are in our own way, and whether we know it or not, complicit in this.

Because when the Commission sometimes makes a proposal and then they are getting bad vibes from certain Member states, they are not necessarily the leaders, but from the people in those Member states, from us in those Member states.

So I think it's important to do that as well. I also think it's really important to be honest. Whatever difficulties, whatever institution is facing, you have to be honest about it.

And when we began our investigation into the sinking of the Adriana in June, when over 500 people were drowned, we decided to do it for the very simple reason, just to answer a simple question. How is it possible that 500 people can die like that in European waters with loads of people looking on? How does that happen? I want to know myself how it happens.

And specifically, because we can't look at the Greek authorities, the Greek Ombudsman is looking at that. So we're looking at Frontex, and a few weeks ago we were over in Warsaw, and just trying to sort out, well, what are your responsibilities? It has a search and rescue big element to it.

So what does that mean, a situation like this? So basically what we're trying to do is to show the people who care about this, the relatives of the people who drowned, this is what happened. This is the accountability piece. And I don't think people should shy away from that.

So we're doing that. We're also looking at the EU Tunisia deal, which as you can see, even from the photograph, is highly political. I mean, Mark Rutte, the Prime Minister of the Netherlands, his government effectively fell, partly on a migration issue. Madam Meloni, the Prime Minister of Italy, her whole campaign to win the election, a lot of it was focused on migration. And they were there in the picture with Madam von der Leyen.

So we can't look at the politics, but we are looking at the fundamental rights issues, whether there was a fundamental rights assessment done. If not, why not? In the future, if bad things happen, what are you going to do about that?

Shada Islam

Thank you very much.

Of course, not in a vacuum, the political landscape within the EU and sometimes outside it, of course, plays a part.

I think one of the things that really spoke to me about Sally was when she said the EU staff and the United Nations staff is not really happy themselves with what they are seeing, what they are allowed to do or not to do. And I think that's a very, very potent point that you have made, Sally.

Turning to you now from the Greek ombudsman, just to ask you, of course, you know... This is an issue that is very high up on the Greek government agenda, and I would assume for you too. So if you could please give us a few ideas and some responses to what Sally has said.

Andreas Pottakis

Thank you. Sure. Thank you. Yes, I would like to reflect on some of the things that have been already discussed and heard.

Let me first start by making a distinction that I think all of us should always keep in mind. I heard the figures; I heard the numbers, about 3 million migrants that are legally entering and being absorbed within the European Union annually. But we have to always keep in mind that there is a distinction between migrants, and refugees and asylum seekers. While when it comes to migrants, there is the possibility of regulating the numbers of those that we can absorb and accommodate within the European Union, the same does not apply with refugees. When it comes to refugees or asylum seekers, we have an obligation under International law to protect them. It is an entirely different approach.

Now, the second point I would like to tackle from what has already been raised, I have heard the word vacuum several times already. And I do agree with the member of the European Parliament, with Professor Strik, that we are not operating in a legal vacuum. But I am really, really concerned that we are operating in a political vacuum here.

And while I have to admit to Emily how honoured and grateful I am for having been invited several times to address the ENO on the specific subject matter, I am also a bit depressed that I have to address the ENO on the same subject matter in the course of the last five, six years.

And this demonstrates to me that still, within the European Union at least, the policies that are being implemented are reactive instead of proactive. We are still lacking a comprehensive strategy, a comprehensive plan on how to deal with the phenomenon. It is not a crisis, it is a phenomenon. It will not go away.

A third point I would like to make on what has already been mentioned, and I think Sally has already given us some figures.

We do know, we do recognise that there is an increased need for improving and enhancing the capacity of Frontex. We have been told, I think the figures demonstrate an eightfold increase in the budget of Frontex since 2015. I am not arguing that whether this is a good or a bad thing.

But I would like to raise the point, well, to what extent has the European Union been equally sensitised and sensitive to the need for investing funds on integrating those that have entered the European Union? To a great extent, to a large extent, and this is part of what national ombudsmen have to deal with, policies on including, on inclusion, on integrating third country nationals that have entered the European Union have been left to member states.

And the involvement or the assistance on the part of the European Union has been minimal, if I may put it this way.

So how much funds have been invested within the European Union on this specific aspect of handling third country nationals in Europe? I am afraid I have to admit that not many.

Now, what can be done? And may I finish with this?

Certainly, things, and again I will agree with Professor Strik, there are no easy solutions or no easy answers to the problems that we all have to face.

However, by avoiding the problems, avoiding reality, we are never going to reach any conclusions that could serve the purpose of trying to find a solution. I am not suggesting that we will find a solution easily. I am afraid, though, from what I have been hearing and from the developments at the European Union level, that we are discussing a migration and asylum pact for seven years already.

Now we are expediting this discussion because, why? Because we are afraid that the new European Parliament, let's be frank, because we are afraid that the new European Parliament might probably be slightly more populist or less receptive to a migration and asylum pact that might serve both the interests of Member states and of asylum seekers. So we are expediting this process within the European Union. You may correct me if I am wrong.

But this is not, clearly this is not the approach to the problem. This is not the way to find, to seek solutions.

Shada Islam

Andreas, thank you very much.

So two questions that will go to Monique for afterwards to respond before I turn the floor to some of you.

Going to you now, Reinier, because of course migration, refugees, asylum is a big, big politically explosive issue in the Netherlands and you have got elections coming up as well. So I was just wondering how would you react to this as an ombudsman. I mean, from your point of view as an ombudsman, what is your role when you hear this and how do you manage this tension between values and political reality?

Reinier van Zutphen

Well, thank you for that question. How to stay out of politics and being political at the same time. That is the question you posed me.

I would like to go back to what has been told. It is about how do we meet and where do we meet the people that it is all about. Go there, meet them, see what is happening. Why are they there? What made them move? Why did they leave their countries? I think one of the myths is that people would like to leave their countries. They do not want to leave their countries. They feel forced to leave their countries, you know, given the circumstances, or family reasons or threats or whatever. And they come also to the Netherlands.

And the number of people that came to the Netherlands this year is I think some 30,000 less than last year. But the problems we have, have grown.

So what is happening in my country, and I don't think it's only in the Netherlands that this is happening, it's about solidarity within our community, in our country. When you are an asylum seeker, (when) you would like to have residence in the Netherlands, or a permanent stay. There is only one place you can go to. That is in the north part of the country. And there we have many hundreds of people applying for a status and first registration. And we do not have enough beds. And we are below the standards that our conventions and human rights, etc. All the necessities to receive people in a proper way are not fulfilled over there.

And if we ask communities in the Southern part or the middle part of the country, would you please help us out and give us beds for shelter? And you know, there is no sort of... Nobody wants to do that. And if we would like to have a law which said, well, we can probably spread these people. It is a very difficult political process; we still do not have that law. And if I go there, I think that's what most ombudsman would do, you have to go there.

As Sally has it fully right, we have to go there and see how we can give a voice to those voiceless people. And not by expressing their voices by myself, but they should be heard. They should be in a position to tell what is going on in their lives.

And if we do that, I hope and think that some of the things that are absolutely mistaken in the public opinion in the country of the Netherlands... For example, that there are about tens of thousands of migrants asylum seekers... Only 10% of all is asylum seeking, you know, refugees, all the others are for other reasons moving to the country.

For example, now we see people (coming) from Poland, Ukrainians who got shelter there, but know that the way if sort of accumulates or let's say, come to the Netherlands... The idea is that you might be better off in the Netherlands than you are in Poland. So they move from Poland to the Netherlands for all sorts of reasons.

In the political debate, it is what you ask for... And I have to be careful here because, you know, I'm not a politician and I would like to have my status unshattered when I go back.

If we are not able as ombudsmen, or politicians, or people who are leading the country... to tell what is really the case, what's really the numbers, what's really going on, why they are here, what our own interest is in having people migrating... Because they are easily workforce for us, in certain circumstances, that we would not like to do the work ourselves. So if we do not establish the facts in a proper way and tell them that there are real facts, and from there on go on and see who needs really our support and shelter, and what are the normal standards that we are now below of.

If we do not do that, we will be back here next year and the year after, and even seven more years as Andreas already pointed out. I was here seven years ago also. It is more or less the same ideas, intentions and experiences. I think the only way to do it, to solve the problem is go to them, make them tell their stories. And yes, as an ombudsman, say: “you complained. Yes, you're right, but I do not know what to do”. It is the wrong answer. But we can't give the right answer for that. We need other people. So how to involve others in what we see as just.

Was it too political?

Shada Islam

No, you have been absolutely honest about the dilemmas you face, Reinier. And I think Andreas has as well. I mean, we are in a political Union. We are in a situation where there is lack of solidarity between, among Member states, where there is tension between our values and what we are actually doing. I think we are all realistic enough to recognize that. I think Sally's point is to humanize the conversation and to make us feel the pain, and to understand that there are perhaps alternatives.

So I am going to turn to you now, Monique. You heard the questions. One of the questions, of course, that Tineke has asked also is about the legality of these deals, these “Team Europe” so-called deals that we are doing. How do we ensure that fundamental rights are actually part of these deals and aligned with our values? Can we do that?

And the second is Andreas’ questions about integration and about also how we have to actually make sure that the money we're spending on Frontex, on keeping people away, there is some kind of a proportionality about how people are when they are inside. What are we doing to make them part of our landscapes?

Thank you.

Monique Pariat

Thank you. There are many questions and of course many things on which I would like to react.

On the agreements, on the Turkey... We could start with the Turkey statement already, it was four years ago. On the Tunisia agreement, it is not quite an agreement, but a memorandum of understanding, yes. We have obviously, every time we negotiate with a Third country, we remind that we have to abide to our legal framework and to our fundamental rights. But the difficulty is also that we deal with sovereign states also that have a different legal framework. So we include in any case a request on fundamental rights. The money we give them is subject also to the respect of fundamental rights. The money we provide to countries are now, and in particular the money that we in DG Home, we are part of the... I don't want to get into technicalities, but we are subject to what we call the enabling conditions, which is there has to be a respect for the Charter of fundamental rights.

So this is verified and we have to do this check before we agree on the programs that Member states will put in place to support their migration and asylum policy and their border policy. And this is part of what we check at the beginning to adopt the programs, and we have one or two Member states which have not been considered as compliant.

So it means we can still launch the program, but we will not pay as long as this is not done. So that's one of the ring fence we have now for this period. And that goes for also the Agencies and everything.

I would like to react on two points that Andreas made.

On the pact, you said we need a comprehensive framework and I agree with you and this is what the pact is about. We have been negotiating, as you say, the previous Commission and this Commission. The previous Commission failed, we started again with the migration pact, we've been negotiating for four years.

So I don't quite think we are expediting the pact, but we know if we don't get an agreement now, I don't think we will get one. So because we will have indeed a new Commission, a new Parliament... This will delay the whole thing for one year... And I think it's a hard time after eight years, even more than eight years, that we finally have a comprehensive European migration and asylum framework that will set a more predictable and a better management of migration.

On the inclusion of refugees, here again, and I agree with you, there is a distinction between a migrant and a refugee. The right of asylum applies to refugees. What we have now is a lot of people arrive, not because they are asylum seekers, but because they are economic migrants with a legitimate expectation to have a better future... But they come and ask for asylum because it's often the way they know we will have to examine and that's also something I would like to repeat.

Every person that is asking for asylum has the right to have its claim examined and this is what we do, what is done in the EU. But of course, many of them don't fulfil the requirements and need to be returned to their countries of origin.

But for those who are accepted, they fall under the Inclusion policy that is covered by the European Social Fund, and the European Social Fund has much more money than the Fund for Migration.

So it's also up to the Member States to use the European Social Fund to promote and to facilitate the inclusion of refugees and this is what we are doing actually together with my colleague in employment, also for the Ukrainian refugees.

Shada Islam

Thank you, Monique.

I am going to now thank you very much for those clarifications. Let me turn now to the floor and I would like to give the floor to a few people for comments and very, very short questions. So please a show of hands and introduce yourself.

I think we will begin with Marino Fardelli, Ombudsman of Lazio. Short comment, please.

Marino Fardelli

Yes, thank you. You have always heard it and it has been repeated as a refrain, Italy must not be left alone. You have always heard it and it has been repeated as a refrain, Italy must not be left alone.

A refrain that you have heard both at the Commission and at the Parliament, but Italy is alone and abandoned on the subject of immigration. We have also heard today that this agreement with Albania is not well seen.

So in our work we are always often committed to restore dignity, the word is dignity, to those who address our office. What is Europe doing to really restore the word dignity, to use the word dignity on this concept, on this subject that we often repeat, often...

Is Europe ready to change its attitude on this phenomenon that unfortunately leaves Italy alone? And if this agreement with Albania is not good, as it has been declared by someone recently, it is perhaps because it is a shy sign, an intervention in a context of a political void of the Parliament and the Commission, which have often addressed these issues, but which have not found a solution.

So what should Italy do? Should it still be with Europe, which is turning the other way? Or is there a concrete commitment, and not just to come to Italy to appear in the pictures, due to the proximity of the next European elections? Thank you.

Shada Islam

Thank you very much.

Yes, Tineke, why don't you respond quickly now?

Tineke Strik

Thank you very much for this remark and for this urgent call. And I fully, fully agree with you that actually what Reinier said about within the Netherlands, there is a lack of solidarity. We really have that big problem in the European Union as well. And I think we do leave Italy, but also Greece and other countries at the external borders, far too much alone. I visited Italy not so long ago and I also saw the problems... But I was also very concerned about the new laws that I heard, you know, that are being developed in order to cope with it.

And for me, it's really key that there will be another distribution system of asylum seekers within the European Union. Otherwise, we cannot solve the problem of non-compliance, of non-enforcement. That is the first thing.

And this is why I was very much disappointed, to be honest, with the proposals on the Pact. Because I really think you need a good analysis.

What is the problem that we have with the European Union at the moment? Because we have our directives, we have regulations, not that we do not have anything, but we lack solidarity. And this is because of the Dublin regulation, of course, having the rule that the country of first entrance has the responsibility for processing or for protecting or returning people. And despite the big resistance of especially the Med-5, it's still in the new proposals as well. So apparently, there is reluctance to depart from this principle.

And I think that has really led to a lot of tricks or creativeness of members of countries at the external borders to evade their responsibility with pushbacks... With Albania or Tunisia deals or whatever, and other Member states who look away because they know if they start to address that, countries like Italy and Greece will say, OK, then you have them.

So this is, I think, the main problem that we are facing that is not solved by the new Pact, because there is mandatory solidarity, but Member states are still allowed to simply pay their obligations off. And I think that's the biggest problem.

On the contrary, countries at the external borders will get more responsibilities because of the border procedures that have to be, people have to be processed at the external borders in detention. So I am really afraid that this package will not solve it.

And we need more political willingness to really tackle the issue of solidarity in Member states themselves, so among municipalities in the EU, but also worldwide, because this whole externalization tendency that the EU is also fuelling only makes that we shift the responsibility to the fragile and poor countries.

Thank you very much.

Shada Islam

Lady over here, please.

Renate Weber

Thank you very much. Renate Weber, the Ombudsperson from Romania.

I used to be a member of the European Parliament for many years, and I remember having very similar discussions in 2012, 2014, 2017, all the time. And I am very sad to say in 2023 that I repeat what I said also during those years, that the European Union cannot claim to be recognized as an international actor, as long as we are not able to have a coherent policy on migration and asylum.

And yes, we do have now something on paper, but I am talking about policy, which means also how to integrate those coming here, how to show that we have solidarity among ourselves.

So this is sad to say, but this is where the European Union has been for years. And I have a question. I used to be a member also on the Civil Liberties Committee, but also on the Social Affairs. And to be very honest, I make this distinguish between refugees and migrants, or migration and asylum seeking. And those who leave a country for various reasons, but war zones, I don't know, police persecution, political persecution, they qualify as asylum seekers. People, I think, most of them coming to Europe, they come because of economic, social, big, huge problems in their own countries.

But it is also true that we encourage them somehow to come by saying and repeating this again and again, that we lack labor force, because the population in the European Union is aging and we need people to work here. Now my question is, now we have digitalization, robots, artificial intelligence. Is this still the case? Or we are selling dreams to those people and they come here and then of course they are delusional.

Shada Islam

Robots have taken over. Okay, thank you very much indeed. A question I will put to all of you in a second. Any other people coming in? Yes. I see you. Yes, please.

Javier Hernandez

Good afternoon. I have been checking the reception centers of the so-called Atlantic Way for weeks. Here are my colleagues from the Canary Islands.

The Mediterranean has problems, but the Atlantic Way, especially because of arrivals from Senegal, is becoming a real problem in Spain, which internally is also causing problems, because it is in the Spanish system of different autonomous communities, although the Canary Island colleagues are the ones who suffer the first shock. We are, the rest of the autonomous communities, those who have to assume their shelter. The Greek colleague, or the colleague from the Netherlands, said... that what they want, and I talk a lot with them, is that they want only one thing, to work and live a little like us.

And the question I would like to ask, if someone can answer it, I have asked myself for a long time. I will never understand why the political persecution of another nature requires a first-class status, and starving in one’s own country does not. I leave the question to you.

Shada Islam

I indeed ask the panellists to take up your question.

I saw a hand go up there, please. Yes, and up there. So, three hands coming up. All right. Yeah, oh, four. Okay. Very quick questions, please, so I can go back to the panel. Yes, please.

Joseph Zammit McKeon

Joseph Zammitt McKeon, Ombudsman for Malta. How close are we for a final, conclusive revision of the Dublin Treaty?

Shada Islam

Thank you very much indeed.

There was a question over here. Yes, please.

Dmytro Lubinets

Thank you very much. I am Dmytro Lubinets, Ukrainian Parliamentary Commissioner for Human Rights, Ombudsman of Ukraine.

First of all, thank you very much, dear colleagues, for the position of solidarity with Ukraine and with Ukrainian people. You know, inside our country we have 5 million IDPs, about more than 6 million refugees, now Ukrainians, in the European countries. This is the biggest transfer of population after the Second World War ended. And all this time, European Union and countries have supported Ukrainian people.

But what do you think about new initiative when governments of different European countries try to establish a new program? And this program, it means that they want to pay money (to) Ukrainians for returning back to Ukraine. Maybe it will be better to send this money to Ukraine and at first create conditions of safety and rebuilding and winning of this war, and only after that we will provide the similar program. Thank you.

Shada Islam

Thank you very much.

Very quickly, please.

Citizen

Yeah, I just want to... I am just coming at this as a citizen and as someone who works here as an administrator. I have no political connections or influence.

I just want to ask if people in politics and in the Commission realize the effect that all of this has on people who are well disposed or even pro-EU and when they see the clear double standard in how human rights, in how human dignity, with Ukraine and Gaza, war crimes, things like that. The double standard, how these people who are allies, who are people who believe in the EU and think it is a good thing, are becoming so disillusioned. I honestly find it so hard to come into work these days.

Anyway, I cannot put it into words what I am trying to say, but I hope you get the message.

Shada Islam

Thank you. We do indeed get the message. Thank you very much indeed.

I saw a hand go up there. All right, please, quickly, and then I am going to go back to the panel. Yes, please.

Mark Dempsey

Good afternoon. Mark Dempsey from Article 19, Freedom of Expression NGO.

I don't work on refugees, migration as a topic, but I have to say coming here and feeling the lack of urgency from those who are supposedly leading on these topics, and I say explicitly, Ms Pariat, I think it is extremely disappointing.

We are coming up to an election cycle. This is an extremely important question at a panel which is prominent and many are watching, and yet the lack of urgency, the lack of research into your own speech was incredibly disheartening. You and your DG are supposed to be showing leadership on this issue, and there is simply none. This has been a topic, as the previous speaker has said, for many years, and all you are doing is signing agreements with countries outside of the EU, which flagrantly go against the Charter of Human Rights and other human rights legislation. So it just shows the pure lack of urgency around this topic.

Shada Islam

Thank you for that very pertinent question.

Can I see one more hand go up? I don't want to say, yes, I do see you, so perhaps very quickly, like the others have done so far. Thank you.

Petra Fretter

In short, my name is Petra Fretter. I am a member of the Saarland Landtag, Germany, European politician and chairwoman of the Committee on Petitions and I have to say that it has been weighing on my heart for a long time that these same conditions in the EU simply do not correspond to reality. We don't have the same conditions. In my opinion, it is a European equality crisis.

Shada Islam

Just a very quick question now. Thank you.

Petra Fretter

A quick question. Okay, okay. How do you want this perception of the citizens, which is likely to have an impact in the next EU parliamentary elections? We have just mentioned it, towards populism prevention. My focus is on establishing equal conditions here and as politicians it is really difficult for us that the EU, which is so important, you addressed it today, migration policy. I fight for this every day, to communicate this to the citizens and that is extremely difficult if we do not have equal conditions.

Shada Islam

Thank you very much indeed. So I am going to go back to our panellists.

And some of the questions, please don't feel obliged to answer all of them, but I think some of the questions that have to do with population, citizens, and how they are feeling the lack of sense of urgency, how do we avoid falling into the populist trap even further, I would like to add. And of course the double standard issue that has been raised, the question of robots, AI taking over the work that migrants could be doing, and also the more general question of the Dublin Treaty, why are we not? So please pick and choose the questions that you think you can respond to, and I think I will start with you, Andreas, please.

Andreas Pottakis

Okay.

Shada Islam

And also, just if you have one recommendation going forward, put that into your final.

Andreas Pottakis

Sure. Well, I am not going to address whether we are close to reviewing or revising Dublin. I think there are more competent people on the panel to address this specific point.

But I will return to some of the issues that have been raised, and I will conclude with a final note on a recommendation perhaps. It has been mentioned, I have heard about the manner in which the European Union deals with third countries, and it has been mentioned not only the Tunisia Accord, but also I think I have heard about the joint statement with Turkey.

Now, let me remind all of us that the European Union never assumed ownership of the EU-Turkey joint statement. And let me also remind everyone that this statement, the 2016 statement, is the one that is actually regulating basically the condition in the eastern border of Greece, the borders between Greece and Turkey. And as we all know, there is no readmission on the basis of the joint statement since 2020, March 2020. The only people that are being readmitted to Turkey are Turkish citizens.

Now, on the point of whether the vast majority of people that are trying to enter the European Union are economic migrants or asylum seekers. Now, I have an issue here, and my point is that, and especially with the new pact, the one that is being negotiated for so many years, I would not like to see a return to practices like the profiling, the asylum profiling. These are policies that have been applied in the past, and I don't think that they have succeeded. I don't think that they are fully in line with protecting the fundamental rights of those who are seeking protection within the European Union, because they are not based on an individualized assessment of every person. For those who are not fully aware of what profiling means, is that depending on the ethnicity of the person, the ethnicity of the person determines the likelihood of achieving asylum protection within the European Union.

So, these are mechanisms, again, I'm not going to use the word expedite, because apparently it's not very welcome, but perhaps a fast-track process of assessing people who are seeking for some sort of a protection.

And let me finally mention another element of the overall policy on dealing with third-country nationals. I think it has been mentioned as well, returns. It is a failed policy within the European Union. We all know that the vast majority of people that are being returned have nothing to do with the people that have entered the European Union, or have attempted to enter the European Union at least since 2015. 75 to 80 percent throughout the European Union of people that are being returned are Albanians. They are not coming from sub-Saharan Africa, or North Africa, or Asia.

My recommendation. Having said all this, I think it might sound a bit paradoxical, but I think the solution is, and will always remain, in more Europe. The solution is Europe. The solution is, and can only be European, pan-European. I am urging everyone to consider comprehensive strategies and comprehensive policies at the European Union level, so that we can all tackle the problems, all these issues, more effectively. Thank you.

Shada Islam

Thank you, Andreas. Tineke, please.

Tineke Strik

Thank you. Thanks for all these pertinent questions.

Maybe about this populist trend to start with first, how to reverse this or to stop this. That is really difficult, but this is really a question that also politicians should think about much more. Actually, to be honest, I see almost the opposite, that migration is seen as a business model for politicians, that they use it creating a crisis or increasing the fear and then say, but I have a solution, namely fences and stopping the migration. This is not in line with our values, but it's also not effective at all.

What I miss is an honest story of politicians telling about the challenges, but also the possibilities that we have, if we do it in an equal, responsible way. We can deal with hosting refugees worldwide, Europe-wide, national-wide, if we cooperate together. I think you should not create the illusion that you can stop migration completely, and get rid of everyone in your country. That would be the responsibility I would say that every politician has, but it's very difficult to make politicians, all politicians act in that way.

Then I come also to the AMMR, because this is the successor of Dublin. There I think it's important that politicians also say in their own country, we need to have an equal responsibility system and we can do that and that will only lead to more predictability and better way to receive asylum seekers.

I saw Mrs. Pariat say, no, no, no, not true, when I talked about the lack of solidarity in AMMR. It's true, there is a bit more than now, there is mandatory solidarity, but there is no mandatory location. I think that this is really the problem, because this is the one that countries at the external border need and that the other countries don't want to give. Therefore, I see the gap emerging. We are now in the final stage of Trilogues and the Presidency of Spain is very eager to get it finalized before Christmas. I'm a bit sceptical about that ambition, because there is still a huge gap between the Council and the Parliament on all the files. But I also see a huge political pressure to get the deal done and maybe then not before Christmas, but in January, February.

My concern is actually that it does not go at the cost of careful and thorough good legislation, because in the end you will all be faced with, if we don't do this in the right way, with a lot of deficiencies that we will find out in the end in the practice. Especially also if I think of the crisis regulation, where Member States are allowed to derogate from all the standards, that will create a lot of national differences, but also really huge tensions with the standards and the values that we have in the European Union. So you will need to work on that a lot.

On double standards, I fully agree with you, I also feel the same. If we do not practice what we preach about universal rights, then we become less credible. Let's be honest, we are less credible for our citizens and also in the world wide, third countries will say, look at what you do yourself. So I think this is a big threat for the EU.

May I say one thing about the Ukrainians, because I really understand your concern? I think it would be really worthwhile to think ahead in time about the future of beneficiaries of temporary protection. Of course we hope there will be a long lasting peace within the three years, but if not, we need to be ready for if not. So that they are not forced to go into asylum system, but then that they get for instance the free movement rights or whatever. I do think it makes sense to offer people the possibility to go back and see if they can build a future, but to always have the possibility to turn back again to the European Union. So not in a way to make them leave, but just to give them more freedom, more mobility to find out what their situation is in the country. I think that's the only way forward on a voluntary basis.

Shada Islam

Thank you very much.

Monique, please. Reinier and Emily then.

Monique Pariat

Thank you very much.

Maybe to follow on that, this is indeed part of the overall negotiations with Ukraine that are taking place. We have prolonged the temporary protection coverage until spring 2025, as you know, but this doesn't mean that people should stay here if they want to go back. What we need to ensure, as Mrs. Strik was saying, is that they have the possibility to go back if they want or to stay. I am sure in the context also of the accession negotiations, this will be discussed. So that's certainly one thing.

On the solidarity more globally and to reply to the fact that the Ombudsman of Italy is saying “Italia é lasciata sola,” I don't think this is true. I think we have supported Italy as we try to support all countries with funding, with support, with the agencies. This is what we are also trying to do as much as possible, and that was demonstrated with the presence of the President and my Commissioner in Lampedusa recently. We are supporting financially to take people from the island, from Lampedusa to other reception centres, constructing new centres.

And on this agreement with Albania, we are looking into it. We need to see for the moment, we have to assess. I don't want to make any judgement on it for the moment because we need to have a thorough look at it. We were not informed about it, so we learned it two days ago, like everyone. So we have to look at it, to look in detail what it will be, and we will discuss it with the Italian authorities and see how compatible it is with EU law.

One thing I would like also to add is on the burden sharing also, because yes, clearly the Member States of first entry share the burden of the irregular arrivals.

But I would like to give you two figures because I think at times we need also to come back to evidence. We had last year, as I said, 300,000 more or less, it is broad figures, regular entries into the EU. Let's say half of them were in Italy.

We had nearly a million of asylum requests. Most of them are, the first countries where asylum requests are made are Germany, France and Spain. So it means, these two figures mean a large majority of people arrive to the EU by normal ways. They are visa overstayers or people that enter visa-free, and not people that come irregularly, including on search and rescue operations. So I think it shows also that the burden is not always where we think it is.

And on Dublin revision, because Dublin is very much criticised, I understand that and it's far from perfect. The Pact is trying to correct that. But we have the choice between two possibilities, to put it simply. Dublin says by default the countries of first entry is responsible. By default. We have tried in the Pact to extend the notion of families to the siblings or to people that have studied in a country. This has not been accepted by the Council.

So this is part of the negotiation between the Council and the Parliament, because the Parliament kept it. When the gentleman was saying that we have no sense of urgency, we put the Pact on the table in September 2020, when the Commission started in November 2019. The negotiations are now between the Parliament and the Council. The Commission has done its work four years ago.

And lastly, to continue on Dublin and to finish on that, the option is either the country of first entry is by default responsible or migrants can choose wherever they want to go. Which is the other option, if we don't have another system. So that's the other option.

So you can understand that there are a number of Member States that are not particularly favourable to this option either. So this is the political landscape we are discussing. All these extremely complicated issues about migration.

Shada Islam

OK. I am sorry, Monique. We are a little bit over time, actually three minutes over time.

And I still have to give, but thank you very much, give the floor to Reinier for a minute and then Emily for a minute as well.

Reinier van Zutphen

Why me again? I will do it staccato.

Shada Islam

You can do it.

Reinier van Zutphen

I can do it staccato. First of all, urgency. The questioner is leaving, has already left. But I would like to answer that it's not a lack of urgency.

It's a sort of paralysis we are in. So we do not know what to do. It's there for four years and nobody is reacting in a proper way so that we can take and get to solutions. It's high time we do that. So we should un-paralyse. That's the first step I would say that is important.

Something about populists. Take populists very seriously and from time to time think about yourself. Do I have some populist elements in my own reading, reasoning and things that I do, (it) might be very important to realise what people really feel when they are living in a country where... etc, etc.

It is all about human dignity. We should give people that come to our countries and are allowed to stay; we should have them at work as soon as possible. We should give children education as soon as possible. We should integrate them as soon as possible. We should not put them aside, as long as they do it in my country.

And a general remark. This is a bit out of my scope but, I think that when we are talking about that there is only external boundaries in the EU then we should take all the consequences of that position. And we do not do that. It is half-hearted. We say there is external borders but we do not accept the consequences for what that means for all of us in accepting the number of refugees and asylum seekers, etc, etc. So if we say, and even my king has said that in speeches, there is only external borders in Europe, then we have all the consequences to face. What is happening next in our Member states and countries.

And then, last because I have only one minute. I go back to your book and what you said to us. And I think four or eight years ago the great gathering of ombudsman, their conference, was titled “Giving a voice to the voiceless”. And I think every morning I wake up and I hope, every morning all the people that work in my office wake up, is that, let's do that. And if we do that, I think all the policies we have been talking about probably will be realized in some years. But giving voice to those who need a voice. That is the most important part. It's all I repeat myself. It's all about human dignity.

Shada Islam

Thank you. Thank you very much indeed, Reinier. So Emily, a quick reaction from you.

Emily O’Reilly

I think the one word that hasn't been used today, although this afternoon, although Andreas did reference it when he was talking about ethnicity, and that word is racism. And I think we have to acknowledge the extent to which racism sips into the thinking behind the development of policies when it comes to migration.

And Madam Trik has talked about honesty. I spoke about honesty myself earlier. So a bit of honesty wouldn't go astray in relation to this. I think in relation to the to the migration pact, I know there is optimism about it. I know it's intended to have that balance between the responsibilities of the receiving countries very often Italy, Greece and so on. And those countries, the other countries.

But, in relation to burden sharing, but it doesn't mean that countries actually have to take refugees and migrants. And my own country, Ireland has already said “we're going to pay the money. Not accept”. So again, we have to wait to see how that how that pans out.

I see, you know, OK, perhaps positive improvements in it, but unless there is that genuine solidarity, we're not going to see that improvement.

In relation to my own work. I mean, you know what I said about Sally's work is that it shows us things that we cannot avoid. It holds that, that mirror up to us in relation to what is actually happening to human beings. So it is in that sense being honest about the reality behind all the policies, and protocols, and treaties and pacts and so on. What actually happens on a daily basis to human beings. And I would hope that the work that my office is doing, your office is doing, in my office particularly now in relation to looking into what happened when 500 people died in plain sight, in European waters back in June.

And the work we're also doing in relation to looking at the fundamental rights issue that that will also be there as something for people to look at and observe, and draw from when they're making future policies.

Thank you.

Shada Islam

Thank you very much Emily. I am not going to wrap up this discussion, if you want to know what I think I write about this very, very often in EU Observer and often in The Guardian. So I will spare you my thoughts, but thank you very much indeed to the panellists for their very, very insightful thoughts and to Sally above all for bringing honesty and dignity into the conversation here in the European Parliament.

So now, we have time for coffee and tea. We are a little bit late nine minutes over time. But I didn't want to stop all this conversation which was very, very interesting. So please come back to your seats. Help yourselves to the refreshments and come back here at 16.30, 4.30 when we start our panel on artificial intelligence and rights as well.

Once again thank you to the panellists. Let's give them all a big hand. And to Sally.

Day 1 Session 2 Lessons from public administrations using AI

Shada Islam

Hello people. Please take your seats. I'm back. We're back. We're going to start our second keynote and then a panel. Please take your seats.

Hi, Michael.

Please take your seats, everyone. Could I ask you all, to please sit down? Thank you. I can understand there's a lot to discuss and it's great to see each other in person.

Thank you very much indeed. Okay.

So we're going to start our second session and as promised, it's about artificial intelligence. And of course, we've already sort of hinted at it in our discussion on migration. Artificial intelligence is taking over, developing at great speed and raising very, very important, profound questions about how we will live and work in the future, actually, even now. And so there is a lot of talk here in Brussels and other capitals. London just had a big summit on regulation. Can we regulate? Does it have to be voluntary? Will governments be involved? Will it be left to business? The main thing is, I think for all of us here, can we really ensure that artificial intelligence is used in a way that does not undermine fundamental human rights, human rights of democracies as a whole? And to hear his point of view, very, very happy to invite Michael O'Flaherty, Director of the EU Agency for Fundamental Rights, to give us some remarks on this issue.

Please, Michael, the floor is yours.

Michael O'Flaherty

Shada, thank you very much indeed. Emily, where are you? There you are.

Ombuds people all. It's a great pleasure to be here. I'm delighted to have been invited back. Curiously, the last time I was here, it was to speak on migration. And today I had to stay quiet during that discussion.

But I'm actually going to start by, not to talk about migration, but the fact that I was on Lampedusa a few weeks ago to assess what was going on. But the reason I'm mentioning it to you today is because the people on the ground on the island receiving the people rescued coming off the boats were using AI-assisted technology in a remarkable way.

The Red Cross, which was doing a fantastic job, was working very closely with the representatives of the Ministry of Interior, liaising on a minute-by-minute basis with the Coast Guard using this technology. And as a result, (in) evidently very problematic humanitarian conditions. For instance, it meant through the technology that they could get people off the island of Lampedusa within 48 hours. On some other islands in Europe, it can take six or more months. So that was AI put to the good. I could give a thousand other examples. There's the extent to which our society survived the COVID pandemic in large part because of the role of artificial intelligence. Artificial intelligence for the discovery of vaccines, artificial intelligence for their distribution, and on and on it went.

In another context of the work I do today, looking at human rights and the administration of justice, I'm impressed by the extent, there are limits, but I'm impressed by the extent to which AI-assisted technologies can improve the quality of justice on a day-to-day basis in our courtrooms.

So I'm a great believer in the potential of AI as a force for good in our societies. But I'm not naive. And just like you in your daily work, I also see the risks. I don't have to go to the dystopian end-of-the-world type narrative we're seeing so much in some parts of the media right now. I don't need to go there. I can see in the day-to-day delivery of services, including of public services, of how AI can get it wrong.

I think of the social welfare scandal in the Netherlands in 2021. I think of the problems around the categorisation of unemployed people in Poland. And I commend the Polish ombudsman for the work done to address that issue.

So notwithstanding the great achievements, there's no question that the management of the risk of artificial intelligence means that now is the time to tame it.

How we tame AI has been a close subject of attention for the Fundamental Rights Agency over the past five, six years. We've tried to get beyond the rhetoric to look at the actual risks for human rights and then see how they can be managed in a meaningful manner. That's been, by the way, not so much legal work, but highly empirical stuff.

I'll give you a few examples in just a few moments.

But what I'd like to do now, in essence, in what's left, the time that's left to me, is give you briefly seven elements for contextualising the taming of AI, and then eight elements for the regulation of AI. This is drawn from our work. It's not comprehensive. It's just to give some elements of core issues.

So the context.

The first is something we know from our research.

We have asked thousands of people designing and using AI, why? Why AI? And the dominant answer has to do with efficiency, speed and efficiency, not quality. There's nothing wrong with efficiency, but when it is the primary driver, and if you're in the business of looking after protecting human rights, you have to be very wary, because sometimes speed and efficiency can be at the expense of quality, the expense of respect for human rights.

Second, AI is driven by data, and this then in itself raises all manner of challenges.

In the first place, there's the issue, well, there's across all the data, there's the quality issue, but within quality questions about the data that powers AI, there is the question of error. And when we have looked at data sets being used in AI context, we have found a lot of mistakes.

We have found, for instance, in various AI applications in the context of migration, indeed, we have found errors like categorising children as adults, and with all that would follow on, including through the application of AI assisted interoperability of the system.

So error is a somewhat under-acknowledged problem. And then there's the much better spoken of problem of bias. Now, I don't need to say much about it, it gets all the attention. But what I do have to say is that this problem of bias in the data, therefore generating the discriminatory outcomes, like the social welfare case I mentioned a few moments ago, there's a very limited understanding of the extent of the possibility of bias.

Only a few discriminatory grounds are visible, are prominent to the attention of people who need to know about them. The racial discrimination is there, gender discrimination is there, but many others, disability discrimination, are not well acknowledged in the context of the design and application of technology.

And the technology can't even begin to cope with intersectionality. You deal with intersectionality in your work all the time. The interplay of two features of a human leading to a worse outcome than would otherwise be the case. But AI is only at the beginning of the story of coming to grips with that aspect of bias. All the issues of error and bias are greatly compounded by a distinct feature of the technology, and that is the malign operation of feedback loops, meaning that the mistake gets bigger over time. The mistake grows over time in the application of the technology. And so a problem that is bad at the beginning can become significantly worse at a certain point down the road.

Staying with data, still another problem, is of course the problem of the extent to which we rely on English language data sets, which raises all manner of issues with regard to the quality of the application of technology in other languages.

And the last thing to say about data before I just give you a little illustration from our research is that the errors and the bias we find through, conveyed through the data, can engage every imaginable human right.

If you go to industry, and I've certainly had conversations with leaders in Silicon Valley, and they get privacy, they get non-discrimination, but they don't get much else in terms of how the technology can impact negatively for us across the breadth of our human experience.

Now to illustrate all of this in a single context, we last year, we looked for terms online, which were extremely frequently used. We then took some proprietorial technology for online content moderation, and we fed these phrases into the technology. When we fed into the technology, “I hate Jews”, it worked. The tech flagged that speech as problematic.

But then, when we fed in another phrase that we found commonly expressed online, “I hate Jews’ love”, the technology did not pick it up, because it attached such a weight to the word love that this completely overshadowed the “I hate” dimension of the phrase. So that's an illustration in a way of the interplay of all those elements I've just described to you, and what it can look like in practice.

So that was the second of my areas of context. And I can be briefer on the other ones.

The third is about risk.

Risk is use-specific. The risk will be different for the same piece of technology depending on where you use it. So language, indeed, stay with my previous example, a translation technology is a far more significance in content moderation than it is when you play with Google Translate on your phone. It's the same technology, let's say, but it's the risk, it's the context that gives the risk.

And this is massively important to the extent that we have seen general application AI tech being rolled out to a huge degree in recent years.

The fourth context has to do specifically to the area that you all work in the public sector. And that is that we find in our research, we need to test it again, it's a couple of years ago, but we found that when AI is used or proposed to be used in the public service, there's a degree of trust in it by citizens, which is, let me say, worryingly high.

We give you an example. We asked people a couple of years ago, what would you rather when you're crossing a border? Would you rather your face to be scanned by a machine? Or would you rather the identification be done by a border guard holding your passport? Most people said they trust the machine more than the human. And from what we know about the application of technology today, that's somewhat misplaced trust.

Fifth, and this is only emerging. But science right now, including medical sciences, is telling us to be very careful. Psychosocial problems are emerging from the use of AI. One is, it is having a negative impact on human agency. Because when I go online now, and I go to my social media, I go to the feed from whatever thing I'm looking at, to look at my choices, they're not my choices. They are an algorithm's choices.

And while we may say that and express it and move on, it appears, and doctors are telling us, that they're beginning to see the psychological impact on humans.

Second, we're losing so many spaces for human interaction, because we are talking to a machine. And for many of us, that is just annoying. But for people living on their own who are already vulnerable, that can be devastating. And then to do with all of these, we're seeing increasing references to psychiatrists in the context of the use of AI.

Six of my seven has to do with the need, and this is huge, I can only touch it, the need for us to have an ongoing debate about the limits of the application of AI, the boundaries around AI.

In the first place, we have to ask ourselves, where can AI never be allowed to go? And that's a discussion our societies need to have. Do we have any use, really, for autonomous weapon systems? Is there any basis on earth whereby that is worth the risk and the loss? It's a societal debate.

But one much closer to my work has to do with judicial decision making. Do we now have to decide once and for all that the tech has no role in taking a decision with regard to the well-being of an individual human being? That's because we know that judging, take the role of a judge, judging is not about pure application of logic and the result in a logical outcome.

Since time immemorial, since the origins of justice, we have insisted on qualities like fairness and like equity in the function of judging. And I suggest I defy any tech genius to hardwire fairness and equity into the operation of the technology.

So these are just examples of drawing the limits.

And finally, there's, of course, the vast issue within this context of respectful use of AI in a manner whereby it does not undermine human rights.

And all of that then, of course, leads to a question which is way too vast for me to touch on today, which is the very close nexus between safe AI and democratic rule of Law states. We need the checks and balances of democratic rule of Law states in order to have some confidence in the deployment of AI within those societies.

That's the context for me to talk briefly about regulation. I welcome very much the fact that regulation is getting so much attention right now in the European Union, in the Council of Europe, and as we saw with an executive order from President Biden a few days ago, right now, unexpectedly quickly in the United States.

Now, I don't want to get into what should be in which bit of legislation. The AI regulation is in the trilogues here. I don't want to mess with that. I don't want to interfere in the negotiations in Strasbourg. I've got no business talking about what they're doing in Washington.

But I want to give you eight elements that we feel from our work at the Fundamental Rights Agency are important to be built in to any regulatory context. And I'll be brief.

The first is, echoing what I said already, that the regulations have to have the capacity to address the full range of the risks. Every aspect of human well-being, every aspect of human dignity and thriving, and therefore every human right is at risk through problems in the application of AI.

Secondly, every human right is engaged. And therefore, we have to make sure wherever we regulate that we in no place ever diminish the protections that we have put in place for human well-being. The human rights system, long fought over 75th anniversary of the UDHR this year, hard worked for, is not irrelevant in the AI context. It must be protected. And the levels of protection in AI regulation must be consistent with all that has been achieved.

Third, we have to be careful how we define AI for purposes of regulation. There's a risk of overly narrowly defining it so that very many important and risky applications are excluded.

You know, there are some definitions of AI out there that would exclude the AI-assisted large-scale databases on the European borders. I just recalled what we were talking about a few minutes ago. Do we want them somehow outside the safety net? So the definition is really important.

Secondly, we have to embrace within regulation all the applications of technology.

We can't just exclude national security. We can't just exclude defence. We can't just exclude policing and whatever else it might be. There may be different models of oversight and regulation for different sectors. I accept that. But oversight and regulation is needed for all.

And then the final point is the discussion I hear in some places right now of taking different approaches to the regulation of the public and private sectors are deeply unconvincing for me.

Time doesn't allow us to say why, but I think we have got to, we cannot just have public sector regulations at the expense of the private sector and vice versa.

Fourth, to echo what I said earlier, our regulation, and I believe that the EU initiatives are doing this, has to be able to engage the use-specific applications of the technology. So we can't just have a system that approves the, to stay with the language app, that approves the language app in abstracto. It has to approve it in every one of its uses and applications. And so use case-specific attention, it is really important.

Fifth, remember the feedback loops. We cannot just assess the risk of an AI piece of technology at its moment of creation. We have to have, at least for the high-risk context, we have to have life cycle testing because the mistake can come later. The mistake can, the error can get bigger later, and we have to be able to capture that wherever that might occur.

Sixth, it's obvious we're going to have to have a high degree of self-regulation of AI. It's just too big. It's just too expansive. It's everywhere. It's ubiquitous. And so we cannot have an external regulation of everything. We can manage that through a number of methodologies. The categorization of risk, like is being proposed here in Europe, I think makes sense. You have a much differing degrees of scrutiny depending on the level of risk.

But wherever there is a risk, a high risk, and you have self-assessment, self-regulation, self-oversight, that must be accompanied by strong external oversight. And it's really important, we'd insist, that whatever frameworks for the regulation of AI we have, that when it comes to the areas of high risk, there is strong, well-resourced expert oversight. I use the word expert deliberately because it has to engage all the threats to our well-being, not just privacy, and not just this, not just that, but the whole breadth. And so I'm of the view that you need to have a deep human rights expertise at the heart of the strong, independent oversight of AI.

All of this then adequately resourced.

And when I travel around Europe to capitals and I meet with AI ministers, I'm not sure in many places the penny has fully dropped, sorry, that they have realised, that the sheer extent and investment they're going to need to do the job adequately, I don't think it's been sufficiently grasped in some places.

Seventh of my eight points, we need remedies for violations. When AI, the application of AI does you harm, then we need a basis by which you can get redress. And so we need effective remedies that are visible, are well-known and are accessible. And I mean, you're familiar with this in many contexts, but it's no less important here.

And then the final point, and I've left it to last, because it's so seminal to absolutely everything else I've said about effective regulation, and it's so integral to the function of an ombudsman, and that is transparency.

Oversight of regulation will not work without transparency. Your oversight of artificial intelligence will not work without transparency. But we're getting quite notable pushback around the concept of being transparent. We're told that it's not possible to be fully transparent. We're told it's proprietorial. “We can't tell you.”

We have tested every one of the arguments. We find none of them compelling. There is an adequate degree of transparency that will not violate those principles and is sufficient for purposes of oversight.

And again, we must not yield on this point, because the idea that we would allow secrets in an area with such implications for human well-being, it would be very worrying.

And just before wrapping up, just a brief word on innovation.

We commonly get pushback that the more we insist on a human rights respectful form of regulation of AI, the less innovative it will be. And China will jump ahead of us, and it will have all the fabulous developments, and then we'll be left high and dry. Again, I'm not convinced.

I believe we have really, totally, inadequately invested in sandboxing AI to assess its fundamental rights and human rights compatibility. The sandbox model is a good one. It's used in every other aspect of tech. It's a good one to map the risk without hindering the innovation. You can innovate in your sandbox. You don't hurt anybody in there. And so I'm not convinced.

And by the way, even if it were convincing that we were holding back some kind of innovation, I would say, but look what we're giving you. We're giving you trustworthy AI. And I think in the long run, trustworthy AI is the one that is going to win out.

And so, my friends, just to wrap up, there's so much hype about AI right now. I am fed up reading newspaper articles about it. It's almost like this magic that's floating around and is going to overtake us all and consume our world, and we should just surrender. It's too late. We can't do anything about it. May I suggest to you respectfully that that's complete nonsense.

AI, what is AI? AI is a technology designed by humans for human purposes, overseen by humans.

We can regulate the humans. We can hold them accountable. And people like me, we look to you. We look to the community of ombudspersons of this continent, because you are now already and will play such a central role in delivering on that accountability. Thank you.

Shada Islam

Thank you very much. Thank you very much, Michael, telling us about responsible use of AI, transparency, and of course, also the need for responsible and I think you use the word “taming” AI for ourselves. Thank you very much, indeed, very thoughtful and very insightful and a very good basis for the conversation we are going to have now. Thank you very much.

But before inviting our panellists to the podium, we have a short video that we'd like to share with you. And it's about artificial intelligence, which is already being used in public administration. So let us have a look.

Video

“Public administrations across Europe are already using artificial intelligence to automate tasks and improve efficiency. The EU Agency for the operational management of large scale IT systems is using AI in the biometric matching systems it provides member states for border control.

But one of the dangers of these systems is that they may reinforce racial and gender biases.

The European Space Agency and the European Commission are developing a system in which AI analyses satellite images to monitor farmers’ compliance with subsidy rules. The system automatically verifies land use declarations by utilising databases and satellite images.

The European Intellectual Property Agency registers around 135,000 trademarks and 100,000 designs annually. It is introducing various tools to automate some processes, including providing officials with draft letters to applicants.

The Flemish Agency for Child and Family developed an AI system to predict which day-care services may need to be inspected to ensure child well-being. This system is meant to support agency staff rather than replace their expertise or control. Syri was a system to detect welfare fraud in the Netherlands. It was hoped it could reduce administration costs and make fraud less likely in the first place. But a court ruling ended the use of Syri after finding that the potential damage to privacy outweighed the economic interest of preventing fraud.”

Shada Islam

Thank you very much indeed. I've been told to use the microphone on your left.

Okay, thank you very much, now we start our panel discussion. I'm going to ask our panellists to please come to the stage. Karen Melchior, Member of European Parliament and Rapporteur on Digitalisation and Administrative Law. Max Strotmann, Member of Commissioner Hahn's Cabinet responsible for AI and digital transformation portfolio. Raluca Peica, Director General at the European Court of Justice.

Niels Fenger, Danish Parliamentary Ombudsman and Ülle Madise, Estonian Chancellor of Justice.

Please find your seats. I've been listening to some very important and interesting conversations and AI already being used. Am I using the right microphone, colleagues?

Karen Melchior

I think it is fine. It is mostly when the cameras are on you.

Shada Islam

All right then, thank you for that.

All right, so let's begin with Karen Melchior, Member of European Parliament. First of all, I'd like to just ask your thoughts on Michael O'Flaherty's speech. He made some very important points about transparency, accountability and also the fact that we should have checks and balances. It is already being of course used but can't be used in judicial decision making. But what are the points that you found important?

Karen Melchior

I think it was optimistic to hear that we were not getting it completely wrong when we in the European Parliament were negotiating the AI Act on the committees. Because a lot of the points regarding risk and banned use is also included in what the Parliament came with on the AI Act. The context in which we use AI in the different sectors, what the purposes are and also the good things about AI that it increases efficiency and actually providing decisions in a timely manner. Because as delayed justice is justice denied, and we need to make sure that the decisions are actually given to our citizens.

I think making sure that there is expert oversight is crucial. We usually talk about human oversight but having competent people there and people that are empowered to be able to change the decision that is prompted by the algorithm. I think that is crucial.

One point that I took with me from the speech is also the emphasis on regulating also the public sector. Because often when we regulate in the European Parliament and with the Council, we often end up regulating the private sector. Often also because the Council is not very willing to regulate themselves and put barriers to their use of technology.

I think it is important as the Dutch case on child benefits show it is necessary to also have rules and limits to the use of digitalization and AI also for the public sector.

I also agree with him that there is a lot of talk about AI and we need to, I think, broaden the way that we look at fundamental rights when we talk about AI and not only look at machine learning but also look at digitalization in itself.

I think that it is correct that we look at machine learning and AI in the AI Act but when we look at fundamental rights we should not stop there but look at all uses of digitalization. Thank you.

Shada Islam

There was one thing that stuck with me from what Michael said was also that people trust machines more than humans. Any comment on that?

Karen Melchior

Yes. I mean there is this sketch in a TV series where you have a civil servant saying, “well, the computer says no”. And that is the end of the discussion. And I think we have the trust in machines and in algorithms more than we should have. And we need to be able to question the prompts that we are giving, the suggestions that we are being given so that we can use that expert overview and the human insight into making sure that we have fair decisions and fair administrations.

Shada Islam

Right. Thank you. Thank you for that. We will come back to more points that have been raised, including the whole moral panic that you have referred to as well.

So for Raluca Peica from the European Court of Justice, important question is are you already using, experiencing this artificial intelligence? And what are you seeing? Are you seeing some benefits? And what are the drawbacks that you are seeing from this?

Raluca Peica

Thank you. I mean there is a lot of enthusiasm about this new technology.

I would like to say that at the Court we are prudent enthusiasts. We have taken a risk-based approach. And we believe in what is called governable AI. That means to include a dimension which until now was not included in any technology, the ethical and the human rights dimension.

But also to see that assessing this technology or adopting a tool based on this technology, it's a continuous process.

It was rightly said a few minutes ago that such machine can change behaviour. Why? Because it's learning continuously. And what does it mean? It means that once it is adopted, you still need to revisit the decision that you have taken.

So we believe that red lines are very important and that is the reason for which before adopting anything, we have put in place a governance.

We have put it, introduced it in the governance that we had, another board which is looking for this ethical and human rights dimension. We are having judges around the table to look to this. So we are taking it very, very seriously.

Of course, we see benefits. I would just name three.

Efficiency and effectiveness. This is both for the judiciary process and administrative process. For the judiciary process, we are looking much more on the, not on the decision-making side, but on the preparations of the decision-making. Making the information available, linking information, maybe seeing patterns and so and so.

So I can mention a few applications, detections for references in documents or summaries of documents. We receive cases which are having packages of documents. It is very good to have a summary of one document, or the package, and to have it in different languages.

So we are looking in this area.

Also, to have the transcripts of the hearings much faster than we are having now. We are looking as well to that. It was said a bit earlier, a bit about the celerity of the judiciary process. So we are looking for this type of efficiency in this area.

For the administrative processes, we are having a bit more flexibility there in the sense that we do not have so many red lines.

But however, it is important as well to look carefully in that one.

I will just name one area in which we are using it for the financial processes, the automatic invoices, checks. Then we are looking as well for improving our search engine and having more intelligence adding to it. In the building operations, I do not think it is a secret for anybody. Now with this energy attention, we are looking into the smart buildings possibilities.

Then for the quality and consistency of our judiciary decision, it is very important. We are looking to the automatic correlation and cases classifications. And we believe that these tools will bring in the next years a real benefit on the legal research tools that will evolve.

And last but not least, it will increase the access to justice and transparency. And here we believe that especially for people with disabilities, having automatic transcripts when we are having the hearings ongoing, and this having it in different languages, could be a real benefit. As well as the assistive technology for people with impaired problems. So speech to text or text to speech is going to help a lot.

And I will just name one more.

Besides the translation and interpretation, which we believe that this technology will really put down what we have now, the barrier language in Europe, will be able each one of us to address a question and receive an answer in our own language in a much smoother way.

Also the sentiment analysis will make a change in the communication. Today we are communicating mainly unidirectional. We do not have the input from the audience that we have. And we really look in communication to create these two-ways communication channels. I will stop here, thanks.

Shada Islam

You're seeing more benefits than drawbacks from this? Could you give us an idea of that?

Raluca Peica

Well, balancing the benefits with risk is extremely important. That is why we are prudent.

I will just name a few of the risks that we are seeing and I will explain a bit.

Biases. If a machine, it is built inside, we can control the data with which it is fed. We can assess it regularly.

It's easier to have it. But how many of us has the capacity? I mean, the human knowledge, the infrastructure needed and so on, so to build solutions in-house. And how efficient would that be? Reusing is much better. If we take solutions from outside, we need to make sure that when a machine was created, we are not going to introduce biases, which can be involuntary biases, but can introduce discrimination. So this is an important risk. And that's the reason for which we started first with a clear governance before we launched this way.

Then, sensitive data. In the court, we are having for the judiciary process a lot of sensitive and confidential data. And it is very important to not use a machine which is somewhere in the cloud and which can practically bridge this confidentiality.

Security issues as well. Well, hallucinations is one of the phenomenon that we have, I think, noticed in many cases. It is very important to have a correct feedback. But also, you see how important is the human critical factor. So assessing a result, it's extremely important.

And then explainability. Once you have a machine, you need to understand how it was built up. If you are not able to know that, you will never really fully trust it. And that is extremely difficult to obtain it. And that's why the regulations are important. Because if each constructor of such technology is going to be obliged to accompany the product with an explainability, this is going to help each one of us.

Shada Islam

So you have to actually know your AI. You have to know your robot. Okay. Make friends with it.

Okay. Let's go to Max. So Max, also, from your point of view, the European Commission's plans. And you were sitting and nodding quite often when you heard what Raluca was saying. So let's get your ideas of the drawbacks, but also the benefits that you see. Cool.

Max Strotmann

Thank you. Thank you, Shada. And actually, I thought there must be some natural intelligence that placed me just in the middle of the two of you. Because I was first nodding a lot there and a lot there, which is maybe a bit boring for the audience.

Karen Melchior

We will nod at you.

Max Strotmann

But I tried to find things that you (...).

No, but to pick up something that you said, Karen, that is a little bit under the radar in the discussion on the AI is the role of the public sector and of us as public institutions. And that's what you mentioned, actually, as well.

There was a lot of discussion about regulating, just as we heard from Michael before, and rightly so. And my feeling is that the discussion on the AI Act is going well in the right direction. But that we as public services, as public administrations, see ourselves not only as a regulator, that we stop there, but that we see ourselves as an implementer, as an “experimentator”, and as somebody who can feed back and be open about saying, OK, we ran into a problem. We share why we ran into a problem.

We ask for help of the public sector, other institutions, but also of the private sector and academics to help us unblock that. And this, and that's what you said before, what I liked really a lot, is not only applying for me to AI. AI is such a fancy and nice buzzword.

But Raluca also mentioned it, that this is actually something that applies to all use of technology in the public sector.

And I'm deeply convinced, and I think with me, my colleagues in the Commission, our team around the Commissioner, but also the colleagues across the DGs, is that digital transformation is something that can be co-shaped by the public sector. And we don't do that, or we don't speak about it actively enough.

And for that, we need a lot of cooperation across the public sector, across the EU institutions. Now, you have a board, where you, with Natalia, my colleagues, you are cooperating all the time with the Parliament, everybody else. But why don't we stop there? Why don't we go beyond? And what we are trying to create is actually a positive feedback loop, in that sense, amongst the public sector across Europe, at all levels, at the EU level, but in particular, at the local.

That's what Michael said, what you both said, is that the implementation at the local level is where it really matters. That we learn from the local level that we try to see, if there are good tools, let us share them.

If we have trusted approaches to avoid what you said, sort of perpetuating the problem, let's give it back and say, we have tested it, we have seen it... Give us the feedback if there is a problem.

And then we can sort of build on each other's approaches. But that we are not, (we are) too hesitant in that. Let's be more ambitious. Let's go out and say, yes, we want to co-shape it together with small companies, larger companies, in transparency. I think that's something that is very important for us.

And that's what we try to build, last point, sort of within the Commission, where we pre-empt, we don't pre-empt the AI Act, but we want to sort of have an early implementation start of the AI Act principles within the Commission.

Once we have the AI Act adopted, we can change it.

And the second element is our law on interoperability, cooperation between the public sector, that is very much focused on data exchange, but where we want to create this cooperation framework at all levels across Europe to learn from each other and with each other.

OK, right.

Shada Islam

But you know something, Max, so I teach at the College of Europe, right? And now, of course, all the students are going to be using Chat GPT, open AI, whatever, to write their (papers). So we're already challenging it. And we are still talking about regulation, etcetera, but we are in the midst of it. So I'm just wondering, when you are running, how do you actually then regulate? We are already getting used to certain things. So what are we going to do? We're going to stop and wait for this Act or this regulation to come out? What are we going to do? What should we be doing?

Max Strotmann

Yeah, no, no, good point. I think stopping and waiting for the train to pass and to be run over is not really an option. But, actually, we are all learning from Chat GPT and others, and we are starting to see the issues from it.

So one of the lessons that we see within the Commission, and I guess this is the same for other public sector bodies, is that we are very careful about the data. That we are very careful with whom we are sharing what, that we are trying to build our own models, maybe, to run Chat GPT-type, other kinds of AI kind of applications. And this is, again, something to come back to my point before. I need to do this within the margins of the law. And this is not only the AI, if it comes, hopefully, very soon. But it is also, of course, GDPR and fundamental rights and so on, and so on.

This is Europe's values, and rights, and legal setup. And this has come before. We need to enforce. This is very clear. But we can, again, I think as a public sector body saying, we are transparent. We are applying our own laws. We show you how we do it. You criticize us. You bring us to court if it does not work.

And I think by that, we can push against the negative consequences of possible applications that we see, like in the Dutch example.

Karen Melchior

Yeah, exactly.

Shada Islam

So the Dutch example is there also. But there's also, and I want to bring it to the table for everyone, there's also intersectionality. And it has been clearly proven by some of my colleagues that the algorithms can be extremely racist. And that actually impacts on the refugees, but also brown and black Europeans. So I think those are the issues that we will come back to.

But I'm going to turn now to you, Niels (Fenger). And you're from Denmark. And the public administration in Denmark is already using AI. So I wanted to get some of your experience on that.

Niels Fenger

Well, thank you.

If you want to see how much we use AI, you have to define it, of course. You can define it in many ways. I like Woody Allen's that it is, artificial intelligence is the opposite of natural stupidity.

Then we have quite a lot. But if we use the definition in the draft AI regulation, I think we have around 130 different AI solutions being applied right now. 25% of all Danish bodies use one kind of AI, and many uses more.

The examples fit a bit with what you saw on the video and we've heard already. It could be identifying children who have a need for special public support. We had a program. It has been scrapped now, but it looked basically at how often are the kids to dentists. And if they're not often to dentists, there might be a problem in the family. And that was not used to decide a specific intervention, but simply to find out who the public authorities should talk to.

It could be detecting fraud, profiling who is in the category of fraud. It could be access to documents where an AI machine will find what is covered by the request, identify the documents and point to places where there is sensitive information that might be deleted. It could be assisting doctors in seeing where there is risk for breast cancer. It could be clever buildings, one mentioned it already where you follow the weather forecast to see if you should turn up or down. The weather. And it could be chat robots giving a better service. So we see it all over.

Shada Islam

Yeah.

Yeah, no, while you were talking, I have to say when you said, you know, which kids are not going to the dentist and you talked about, you know, being used for social benefits, et cetera. We have a wonderful, rather terrible example from the Netherlands, where the algorithm was extremely racist. And if you had a Muslim-sounding name like mine, you would then obviously be suspect of having done fraud. So how do you make sure that those guardrails don't go off the rail? How do you make sure that these things don't happen? The ethics of it, if you like.

Niels Fenger

I completely understand. I'm not sure I know the solution, but I can definitely help pointing to the problem. I mean, this example I gave with the children that was scrapped. That was scrapped even before we found out if it worked because there was a public outcry.

But I think there are three ways where you have the risk for bias arriving. We already touched upon it a little bit. But I mean, the first is what we can call garbage in, garbage out. If you have bias already in the practice that you are training the machine with, that bias will, of course, stay in the machine.  Or it could be if you have too little data, then you risk that something which is just a one-off becomes something the machine thinks is a regular occurring thing. So if the data foundations are not large enough, you risk that it goes wrong. The third thing, which we have already touched upon, is this algorithmically enhanced risk where the machine continues to take a given criteria and make it bigger. Let me give an example of some researchers. They were training a model to identify on photos whether it was a man or a woman. And one of the things they did was that the pictures they trained the machine with had a propensity of women in a kitchen. This was a deliberate attempt to show the bias. And indeed what happened, the machine began looking at “kitchen” rather than the face. And it turned out that 84% of every time there was a kitchen, it made the person, or labelled the person as a woman, even if that was not at all the real statistics. And this is the problem.

I mean, we don't know how the machine arrives at the result. Somebody said that the machine is not logical. No, it's not. I mean, it's not based on a logical deduction, a causal logic. It's based on statistical correlations, based on what is trained into the machine. So therefore, of course, it's imperative that you constantly have mechanisms trying to see whether it comes out in a wrong way. The problem is, it's not very transparent how the machine arrives at the result.

And I think it's fair to say, that's at least what I've been reading from those who know something about it, that the more precise the machine is in predicting the right result, the less transparent it is. And that, of course, makes it not only difficult to correct it when it goes wrong, but it also raises a transparency issue. I mean, some of the previous speakers have touched upon it, but I think it is quite fundamental.

We know from most national laws, we know it from the case-law of the Court of Justice, we know it from Article 8 case-law in Strasbourg, that you need to motivate certain decisions, particularly negative decisions.

Now, how do you do that if the machine is a black box making it? You can, of course, say all this is what we fed into it, but it does not answer the particular question. That's not good for the citizen, and that's not good for us ombudsmen or judges who are going to control it later.

We can say, well, we don't care, we just look at the substance, is it a right or wrong decision? But we know where there is discretion involved, that that doesn't bring us all the way. We need to be able to control the decision, and if the decision is not adequate, we might annul it.

That's the problems, I see, the solutions, uh...

Shada Islam

So essentially, human oversight is what you're saying.

Niels Fenger

That's one of the things, yeah.

Shada Islam

So Estonia, Ülle, Estonia was an early adopter, if I'm not mistaken, of AI, and the ombudsman uses AI in his daily work, and is your office already doing it? And what are your experiences?

Ülle Madise

Thank you. First of all, the Legal Chancellor in Estonia is also the ombudsman for Estonia, and very similar to German Petitions Auschuss, just to be clear.

The overall situation currently in Estonia is quite similar to that in Denmark, and all these problems and all these examples described by Niels, they are there, unfortunately. If you add some things, then in Estonia, in public sector, there are approximately 120 services based on AI algorithms, and some of them cause some concern also for our office. For example, the Unemployment Insurance Fund in Estonia uses artificial intelligence as a decision-supporting mechanism, and as you can imagine already, we don't know exactly which data are used to train this algorithm, and therefore we cannot be absolutely sure that the biases are not reproduced, or the mistakes maybe made by some public officials previously are repeated.

If you ask now, what is done in our office: we are quite open to innovation, and of course all the advisors are free to use large language models, including chat GPT if they have to go through a large amount of information, just as support. Of course, all the decisions of the Legal Chancellor, Ombudsman, all our statements, all our investigation visit reports; they are prepared entirely by advisors. And I read them, really read, and I sign. I sign digitally, digital signature, but it is not artificial intelligence. And that's my mind who is looking into these papers.

But our website is at the moment, I would say, old-fashioned, and therefore the new website is coming, under construction, and there we want to build in already several AI solutions. These are already designed. To start from, we want to use all these possibilities that the AI offers to translate the texts, or to convert spoken text to the written text, and vice versa, to make our work better accessible, especially for people with disabilities. For example, for people who don't see. Or if the Estonian version of the Estonian sign language is ready, but also use the subtitles, they don't help always the people who are not able to hear, then of course it will also be integrated.

And it's very important as our role is also to protect the rights of the people with disabilities. Sometimes the public organizations, and I don't talk about the private entities, they forget about easy-to-read language, and it's not only helpful for those people who are intellectually challenged, but it's quite helpful for those who are not fluent in the official language, in Estonian language. We have a lot of Ukrainian refugees, for example, who would like to get the original information, and also other, asylum seekers from Syria, or probably in the future from Pakistan, and so these people also would need clearer text. And of course the translation from the Estonian language to other languages, but there, and now I come to the threats, we have discovered that there is a perfect deepfake technology, what translates my spoken text into whatever language. If the translation goes to the French language, or the Ukrainian language, it is quite easy to check whether it is correct... Or Basque language, we can ask Manuel. But if it's translated into the Chinese (language), for example, or Arabic, I wouldn't have any clue. This is my face, this is my voice, and I am talking. But is the message the same, what I wanted to say? It might not be.

And there I come to the end of my first message. It's very important for us, the Ombuds people, to maintain the credibility. The loss of trust is something you don't win back very easily. Technical errors, personal data leaks, discrimination in our office, by us, or a message in a foreign language, which doesn't comply to the human rights, also possible. And the public trust is gone. And therefore we are way too careful.

Shada Islam

Thank you very much, Ülle, also for bringing up the issue of deepfakes and disinformation, of course, with the elections coming up, in all Member states all the time, but also to the European Parliament. These are important issues to think about.

I'm happy to now open the floor to quick questions, comments. Some fascinating insights have been given, not just by Michael, but by our panellists as well, of their daily experiences, their hopes, their experiments, and also the challenges they're facing. So if you have a comment or a question, please put up your hand. If not, I will continue my conversation. Anyone? Yes, yes, don't be, don’t hesitate, please. Yes, go ahead. Introduce yourself, please.

Adrian

Good afternoon, I'm Adrian Mesuit, I'm here on my personal capacity.

I think indeed the scandal in the Netherlands is a very tragic example. And I know that it was taken into consideration when also designing the AI Act, but also I kind of feel that we only have those references in recitals.

So how strong, can we be sure that everything will be done to prevent further cases? And also maybe if some of the panellists want to, I mean, racism is was mentioned also, well, people with disability on also potential of AI for those people... But maybe talking about potential risks for also that are related to sexual orientation or gender identity, which hasn't been identified yet, maybe it could be interesting to hear from you on that point. Thank you.

Shada Islam

Emily, did you put up your hand? Yes, please.

Emily O’Reilly

Is that on?

I was fascinated with Niels Fenger’s description of the dental check and whether this would need any... and, I mean, I instinctively found it horrifying. And you said, when you started, that this is discontinued, but it was only in your second intervention, you said why. Because there was, there was an outcry. Now, I'm not sure I was that fuzzy about my children's dental treatment, when they were growing up, so, you know, but who thought of this? Whose idea was this? You know, where did it come from? And, and, and were there other similar type of proposed interventions made using that methodology?

Shada

Thank you, Emily. I shared your concern. I also thought, oh, my God, the nanny state is really watching me all the time. But thank you for that.

Any- anyone else coming in? Yes, please.

Tena Simonovic

Tena Simonovic, the ombudswoman of Croatia. So a question and a comment.

First, a comment. While I think it's great that the panel is focusing on the EU level of regulation and the AI Act, I would like to call everybody's attention also to the work happening in parallel within the Council of Europe, mainly the work on the Convention on AI, human rights, democracy and rule of law. And while I do appreciate the effect of EU law and the AI Act, when adopted, will be very important for EU member states, I think this is crucially important, not only because all our EU Member states are also members of the Council of Europe, but also here we are a larger group of ombuds also from non-EU member states.

And finally, because we're focusing on human rights. And it is quite interesting what is happening there as well, which I'm saying also as a member of the, of the group working on the Convention as not representative of my own country, but of the European Commission on racism and intolerance of the Council of Europe. So that would be that.

But my question really is, first, it is great that we are focusing on the issue and there is so much talk about it. But I think it's great that we are also looking at what it means for our work. So I want to thank Ülle for focusing on what she's currently using in her office when it comes to AI.

But one crucial issue, I think for all ombuds institutions, as well as for NHRIs and equality bodies, and all those with multiple mandates is, what will be our role in the oversight? Will we have a role in the oversight? And if not us, then who? Having in mind also what Michael O´Flaherty said, which is the outside expert oversight should entail a deep human rights expertise.

And a further layer to all this is we will have great experts paid by great money in the private sector to design and to have all the explanations. But will we have the right experts in the public sector, working for public sector salaries, who will have the right expertise involving both data scientists, as well as IT experts, human rights experts and equality experts to do that sort of oversight? Thank you.

Shada Islam

Thank you, thank you very much indeed. A very, very important issue raised by the Council of Europe representative.

Anyone else wants to come in at this point? Yes, did you see someone? Oh, yes. Yes, please go ahead.

Ernesto

Hi, my name is Ernesto. I work for Interface, think tank, focus on tech policy.

Thank you so much for your contributions. And I greatly agree that there must be some accountability on related to algorithms, machine learning, maybe something that I wonder if whether what steps are being taken on the fairness on the interaction between humans and machines. Sometimes it's not just the algorithm, it's just whether it is fair that a person has to be dealing with a machine instead of a person, because behind justice, there is this idea of having an interlocutor that listens to your problem.

And I think we all have faced this issue when customer service or apps where we have a problem, and we don't find a way because they don't let us, you know, everything is automated. And our problem is not listed there. And we are just wondering if only we could have someone's telephone or we could talk about our problem. And I think when it relates to refugees, immigrants, this is a very important issue.

Shada Islam

Thank you again, very much for raising that issue. We need to get some people to talk to us once in a while, to answer our questions. Very important.

I'm going to go back to you, the panel. I start with you, Karen, then take Max. And then we'd go this way and then we'd go this way. So please. And I'll give the floor to Emily, of course.

Yes, please.

Karen Melchior

Thank you very much for the input from the audience.

And I think it also shows that when we do digitalization and artificial intelligence in administrations, we sometimes put the cart before the horse because we think about what economic savings we can have, how we can reduce personnel dealing with a caseload, because now we have a digital tool.

But we do that before we've gotten the digital tool working, before we've actually seen that there is going to be the efficiency gains that we are hoping for. And therefore, I think it's important that we make sure that we have the sufficient funding and personnel and experts there when we start doing digitalization. So we actually have a required minimum service level for our citizens.

I think the way research shows that the quality of interactions that citizens have with administrative bodies influences their perception of and the trust in public authorities.

And as a European, a member of the European Parliament and as a rapporteur on a report asking for an administrative law regulating the European institutions, I think we need to look at this and see how this can affect the reputation of the Union. And it's been 12 years that the European Parliament has been asking the European Commission, sorry Max, to put you on the line, to have a law of administrative procedure for the Union.

Because at the moment there are rules and demands on the EU institutions, but it is a patchwork of different regulation. And we know how important it is to have one place where you can see what your rights and obligations are. And I think we need a horizontal piece of legislation in order to ensure this. And we also need, I wrote the report and we've been building on the existing work of the Parliament, including proposals on digitalization that guarantee citizens' rights and consolidating and clarifying interactions with existing legislation.

The end goal is a better experience for those interacting with the European Union's administrations, to have this minimum set of common rights. Because we need to have clear information about their rights in their language and it needs to be correct. That's why we need to control the use of digitalization and artificial intelligence before we hand it out to the citizens. And we need to make sure that they have to only provide the data once.

I mean, what are the good points of having digitalization and artificial intelligence? That could be that you only have to give the information once to an administration so that you have the data shared across the different institutions.

But that also requires that we have transparency and accountability. Because you don't want the information about how often your kids go to the dentist to be spread all across the administrations of all of the land. You need to be able to know that it is used in a factual, necessary, proportional way by the administration. And you need to be able to see who has actually had access to this information.

There was a question from the audience about the AI Act. And I think the point that Michael also had earlier about the life-cycle analysis, so that you don't only look at what is the algorithm like, what is sort of the recipe for the digitalization, but you also look at what does the cake taste like when you've finished, what are the results that come out of the artificial intelligence.

Because we need to make sure that there isn't a negative feedback loop, removing rights from our citizens. I think sometimes we forget the obligations and the rights that we have in other pieces of legislation. So for example, on the gender identity using artificial intelligence to try and identify a person's gender or sexual orientation, if you go back to the proportional use of data, is it proportional to make such an inference of a person's gender identity? And I think in most cases, it will not be proportionate or necessary. And that's why you could already from there, stop such a proposal by saying this is, this is simply not proportional.

And sometimes I think we get too caught up in a sort of AI hype. And we forget some of the fundamentals. And I think this is where the ombudsman's institutions across Europe, and also within the European Union's institutions, and also, of course, the Agency for Fundamental Rights, is crucial to say we have certain fundamental rights, we have the Human Rights Convention, and we need to make sure that this is upheld, even when we look into digital tools.

Sorry for being a bit long.

Shada Islam

No, you have not. Thank you. Thank you very much, indeed.

Max, so obviously answer the questions. But I also wanted to ask you, because there was this summit in the UK just recently, right? And it was all about international coordination. And if you could just say a word or two about what your plans are on that front.

Max Stromann

Yeah, thanks. And this also responds to one of the questions that came up about the Council of Europe.

And there is a lot of work going on in the context of the G7 that you may have seen, where the Commission also fairly, positively responded. The same with a very prominent representation in London recently.

What is interesting that in this context, there is always this discussion, or very often this discussion, the EU goes ahead by regulating, whereas the others sort of are more cautious, or the EU is only good in regulation... But maybe it is, sometimes I think it is not.

And clearly, in this audience, you see a lot of cases where it is not, at least the application of it is not as good as it should be. So, I am not so worried that the EU goes ahead with regulation and that, because I have seen, I have done a lot of work in the past with start-ups and small companies, small innovative companies across Europe.

And for many of them, the reliability of the operational framework, of the legal framework, the clarity, certainty, how I can act and why I can act is super important. And I think that this we overlook a bit.

And some of the comments that I read also in the international arena is that there is a bit of criticism of the very big companies who are positioning themselves very actively and very forcefully, and they have massive, fascinating technologies. But I wouldn't also want to be sort of regulated in this context.

But regulating, putting a frame around that, asking for transparency, asking for explainability, asking for access to the data, asking for an objectability that I can say, no, I want to see this again. This is something that we can force them and that we should force them. And if we do that, I think we will support smaller companies, other actors in the area, also the public sector to, in a positive way, compete with that. So in a sense, I see, and this was also the question, I think there is a lot of parallel work going on.

And in the Commission, if I remember well now, don't hang me, but I think in the official position of the Commission to the G7 and also to the London discussions, it was clear saying that this, we see this very much going in the right direction. And we think that Europe's approach of trying to find a regulatory response that is very measured, but that gives me this reassurance. Yes, somebody will look at the risks. Yes, I can ask for transparency, including (the) role of the Ombuds people. We can ask for that. And I can get, I can have the right to ask for a second opinion and for a human to show me why the decision has been taken and to adopt, object against it, I think is very measured and very positive. And so that's what the Commission President said. We see this going in the same direction. There is a lot of voluntary commitments, which is excellent if it works. But, and our approach, fits very well with it.

If you allow me one thing that came up on skills. This is something that for us, when I said that we are working on our internal AI kind of strategy, that is aligned in purpose on the risk assessment, the “think the purpose before the application”.

Not just because it's fancy, we use it. Why do you use it? And then we come on bias and so on. This goes now, but there is a big chapter also on skills. And I think we're all struggling. I mean, you are a perfect example that the skill is there. In the public sector, there is a lot of skill there, but that we need to continuously learn and build on that, that we need to attract skilled staff.

That is super, super difficult. Also for the, I mean, I can speak for the Commission. It is not evident. We need to all of us learn how to deal with it and not to be too shy and say, hey, stop it. Stop it. It is like when I will go to a counter of an administration and I feel tricked, I feel unfairly treated. I say, stop. Where can I go to complain? What can I go to get a second opinion? And this we need to teach more people, have them understand and be capable for this human part of the public administration.

And that the quality that you refer to, the efficiency is nice, but this is the quality part.

And I do hope, on the positive side, that AI and other digital tools, it's not only a digital transformation helps the public administration to be faster and quite solid on the standard stuff, but to have time for the complicated, the human things.

Shada Islam

So in a sense, AI will inevitably reflect the values and the ethics of the societies that it's operating in, in a sense. So let's keep our values high and let's stick to them. Right.

OK, thank you very much. And moving to Raluca.

Raluca Peica

Thank you very much.

Max touched the point which I believe is very important about the skills and experts. I think we should see this technology. We should not underestimate it. And we should see that it's going to have an impact on the society, but also on each organization. (It) doesn't matter if it's public or private. And we are not only looking for experts. It's true that we don't have it. That is true as well. But it's not looking just for experts. We will need to have strategies that are going to embed the upskilling and reskilling of the workforce, the agility of the workforce. We need to think to all of this. We need as well in the organization to create change-management capabilities. We need to understand that it might be an impact in which a large number of types of work that are done today is going to change. And people need as well to be accompanied in this. So this change-management capability is important.

Innovation is always happening in networks. I believe as well that the cooperation needs to be there.

There was a question a bit earlier, and I am just looking to Ms. Renate Weber, about the robots that will replace humans. They will not really replace it, in my opinion. I guess it's going to change our way of working. We need to be wise to see where we are going to use this technology, how we can leverage on, how can we improve our lives and not just, so not to demolish, but to construct.

However, there is a problem, and I would like to address as well the elephant in the room. These robots that you were talking about, they will be as well migrants, because unfortunately in Europe, we are not yet there. And I think that's another effort that we will need to do. First, we spoke about the data, where the data is going to be placed, who can point the European cloud. We are working on it. We will be there. But I think this is essential.

And then, yes, to encourage the European products to be there as well, because we spoke about expandability. We speak about EU regulation. It's going to be easier if these are going to be as well European. Regulation is very important, but what I want to say is it needs to be as well balanced with strategy. And that is the responsibility of each institution, organization, and so and so.

We need not only to look to the obligations, but also how we want to use it as well to the future.

Shada Islam

Thank you. Niels. Some of the points raised?

Niels Fenger

I will try. If I start with the teeth, and you are allowed to smile when I talk. It was a municipality outside Copenhagen, and what they did was actually quite clever. They looked at the statistical correlation between those where you had a problem in the family, I mean, whatever neglect there could be, and what, and how that had been shown with other factors. And one of them was that these parents didn't take their kids to the dentist, to public care. Another was that they were divorced, and the third one was where they were living. And there were many others. So the logic was actually quite good. What are the risk factors? Where do we see them, sort of laid up? But of course it doesn't look nice, and it was scrapped. And it was also scrapped because the statistical data was so little that they could see that it went wrong.

My second point goes to trust. I think it was very, very wise words that we need to make sure the trust does not go away or is even being even further decreased. I like the, what was it, prudent enthusiasm you said, and I became a little bit scared with the word “experiment” together in the public sector. It depends on where you experiment, doesn't it? If you use it for these types of manoeuvers that Ülle talked about, translating, chatbots, simple stuff, then I do not think the risk is so big.

If you do use it in order to identify where a human should look at a case, which was the case like the municipality, it wasn't the decision, it was just “here you might be able to look at it, here there might be a risk”, then I think it is fine. Then you have the human oversight into it.

And then of course it also depends on the kind of rule. If we have a clear-cut rule where there is no discretion, then I don't see the same risk of putting it on the machine as if you need the human element into it.

Oversight, the Croatian ombudsman. I mean, I tried to point to it a bit myself with saying that if we don't know the criteria that is used by the machine, we have a difficulty in checking that all actually went fine. Especially if there is discretion and it's not a question of a legal or non-legal decision, but has wrong elements come in, illegal elements, or has illegal elements being too much weight or too little weight. Now that's an issue.

What one can do, and we talked about it in Strasbourg, is also to try and force the administration to explain what it's doing. That's an approach we do quite often. We had a new system being introduced where the tax authorities wanted to know the value of your property with a purpose of taxing it. And they made these enormous machines trying to find out what the value was. And suddenly it went, for some people, from say 100,000 euros to 300,000 euros, or 3 million euros. And the minister went out and said, this is absolutely absurd. And it was the computer who made the whole thing.

Now what do I do as an ombudsman? I ask them to show what kind of trials they have been making before they put the system into effect. In all the places where we see that it went wrong, did you actually try to run this hypothetical situation before you put it into place? That could be one way of doing it.

The very last point is, yes, of course, we need regulation. We can ask ourselves if, with regard to the public sector, is it necessarily a European matter or not. It all depends on whether we achieve more by doing it in the European setting than in the national setting.

Shada Islam

Thank you. Thank you very much indeed. So Ulle, let's have your final words, please.

Ülle Madise

Just to continue from the dentist case. In Estonia, there is already in use such kind of service where the municipalities receive data about those young people who are not in education, not working, not in army, and also not at home with the child. And it should make it possible to predict that probably these young people are in need of some help because they are drug addicts or computer games addicts or something like that. And then the people from the municipality come, knock to the door. “Here you are. You are not in education, not working, and so on and so on. Do you need any help? Proactive services.”

Once we discussed it, about it with Rainier (van Zutphen), and now I agree with you. I tried to argue, but now I agree with you that actually in nowadays societies, these kinds of services are needed. And that's where the AI, if it's used properly, can be a very good assistant. But of course, only under the conditions that the human rights are there, followed, and that there is the proper oversight, transparency, etc.

If we come to this detecting of children in need, as a children ombudsman, it would, for me, be very difficult to say that it's not allowed to detect children who possibly need some help because their parents don't bring them to the dentist. Although in Estonia it is free, free from charge, and it is good for the health of the children. And if they are not attending the kindergarten as well, or the school, but we know that they are in Estonia, but something is wrong with that families.

And the third point, I believe that in all countries, certainly in Estonia, in my country, there is a huge pressure to create personalized public services. For example, we have universal child and family benefits. And AI, would it be possible to find these families where the children really need some extra money? Because nowadays, the amount is not so big. And the result is that the universal child benefit is received by families who don't need it, because their standard of life is so high that they just had school holidays in Dubai or something else. And then (there) are other families who need this help, who would need much more. And therefore, maybe it would be wise. And again, under certain conditions.

And to come back, what Tena says that inspired me to emphasize that the need for a proper regulation by the Council of Europe and also the European Union on AI issues is urgent.

And the last remark, please don't let out of the scope the national security offices. I know that they are demanding, that they are left out because the situation is so difficult. But Human rights, with capital letters, please. Thank you.

Shada Islam

Thank you very much, Ülle, for that. Thank you very much for all panellists, to all panellists for a very insightful conversation.

I have to say, I had moments where I was thinking, yes, we are on the right track. And now the moments I was thinking, oh, no, oh, no. Where are the humans in all this? But in any case, I think it increases the responsibility on us, on humans to be actually in the end in control. Big Brother is watching us, I guess. But we have to be careful about what we put in his mind and in his view. So thank you very much indeed for that.

Emily, I'd like you now to come to the podium and give us a little bit of your thoughts, your closing remarks for what has been quite an interesting afternoon, very insightful, very thoughtful. Thank you very much.

(Silence)

Emily O’Reilly

Sorry. Sorry. No problem. I speak now. OK. OK.

Just making the point made by the person from the Council of Europe in relation to how public administrations keep up with the tech companies and keep up with innovation. And I think that is really critical. But, you know, I have somebody mentioned the UK conference in Bletchley Park last week, I think, and... which is very interesting, obviously.

And there was there was a move towards more cooperation, collaboration in relation to safety and regulation and all of that, which is to be welcomed. But I'm still left with the image of Rishi Sunak, the UK Prime Minister, interviewing Elon Musk. And I thought, who's the powerful one there? Who is the really powerful one? Who is really, you know, has the capacity to control and to create in ways that could be good for mankind or person kind or bad for person kind? And I have to say at the moment, I think it's Elon, which is not to be disrespect the UK Prime Minister. But, you know, they know a hell of a lot more than we do, and they have huge capacity for innovation.

It used to be said years ago that the future is rarely futuristic. But I think with AI, it is very, extremely futuristic.

Michael and a couple of others talked about media hype in relation to AI. You know, I listened to some of that. And maybe the impression is given us is this is hype from journalists, from reporters, from general commentators. But a lot of what has been said is by very influential people, important people, knowledgeable people. You know, it is not just somebody spouting off in a column. And therefore, that's why, you know, we do have to have to listen to them.

I mean, I agree with Michael that, and everybody here that, of course, it will obviously produce fabulous benefits. The area that I think, you know, most importantly, will be the area of health and diagnostics and creation of new treatments and pharmaceuticals and so on.

But I think it's also interesting that the conversation has been around how to keep the human in public administration, which is extraordinary, really, when one would have thought that the human has to be absolutely at the heart of public administration. And I have been an ombudsman now for, god, 20 years between Ireland and here. And one of the things I have noticed over time, which has been obviously a time in tandem with the great digital transformation and so on, is that public administrations fall in love with new technologies. They absolutely love them. And that is the point at which the human can be, if not quite excised, but sort of slightly removed from the picture.

And I think when you are talking about what the role of the ombudsman is, it is precisely to keep the human and the person right and centre, front and centre of any public administration. Because we have all experienced just the soulless experience of trying to get through to somebody, whether it is in a private sector or your phone company, whatever, your bank, but also increasingly in public administration.

And the one thing that I also can say, as my experience as an ombudsman, that at the heart of virtually every complaint that we get is an issue of communication and fundamentally of not being heard. And I say to my colleagues all the time, even if we cannot solve a problem for somebody, and this goes for every sort of public service, the fact that somebody has been heard, really heard, not somebody reading off a sheet in a call centre, that they have been really heard is such a stress reliever for people. And we know this in our own lives.

So I don't think we are ever going to get to the point where we are able to hire people with the same technological capacity and capability of Elon Musk and his crew. I think we do have to have people with that expertise. In fact, I have discussed it with some of my colleagues, because we can't just be sitting there passively and wondering what's coming down the tracks.

Transparency, absolutely, it is critical. And we all hear about we need to have the algorithms made transparent. But let me tell you, as European Ombudsman, we struggle with transparency and accountability when it comes to humans. You know, not just machines. And transparency has to be at the heart of everything.

But again, whatever we do in Europe, you know, it has to be in collaboration with all the other great global actors. The idea that, you know, as Michael talked about, the idea that, you know, we all agree not to do evil things, but some country is going to go off and do evil things. I think it is possible that countries will go off and do evil things.

And I think we kid ourselves if we think that we can really control AI, or to understanding or whatever in relation to our understanding of it.

So really, thank you for sparking so many thoughts in relation to that. And also, you know, thank you to Sally for sparking a great conversation about migration. I saw she was having a very animated conversation with Madam Pariat from the Commission afterwards. So I hope that went well for both of you. So thank you, everybody. Thank you, Shada. I will leave you to you to wrap up. Thank you.

Shada Islam

Thank you. So just some housekeeping rules. And these are for the ENO members. So there's a buffet dinner downstairs at 6.30. It will start with a speech by Jean-Marc Lieberrherr. He is president of the Jean Monnet Institute. And the organizers are asking you to meet outside the room where the staff of the European Ombudsman will take you in groups to the dining room on the ground floor. Tomorrow morning, the ENO Conference will start at 9.30. So please be here on time. Actually, registration will start at 8.30. And please keep your Parliament badges so that you can get into the building tomorrow without too much formality.

So with that, have a wonderful evening. Wonderful evening to all of you as well. And let's give ourselves a big hand for all the interesting conversations we have had. Thank you.