Justice and beyond: Council of Europe working on setting global benchmarks on artificial intelligence
Director of the Information Society and Action against Crime Directorate, Council of Europe
JT: The activities of the Information Society and Action Against Crime Directorate comprise standard-setting, monitoring, and cooperation on a wide variety of topics, including artificial intelligence, data protection, action against cybercrime, crime problems, criminal law cooperation, transnational criminal justice, action against terrorism, fighting economic crime and corruption, money laundering and the financing of terrorism.
Although we understand that it is impossible to classify the issues under the Directorate’s mandate by the level of importance or priority, which would you say are the most pressing at this time?
JK: The first one I would say, is to ensure that, during the COVID-19 pandemic, we’re able to fulfil our mission, not just as a Directorate, but as the Council of Europe, as an organisation, that is, to ensure respect for human rights, uphold the Rule of Law, and democracy.
We know that our statutory aims are under pressure because of the pandemic and the measures taken to counter it. We understand that restrictive measures are necessary and that the pandemic imposes difficult choices upon governments. Nevertheless, we should make sure that this virus – which is destroying so much of what is dear to us – doesn’t destroy the very fabric of our societies. For me, this is the overarching challenge that we’re facing at the moment.
It must be clear that any restrictions that are taken, any exceptional measures that are introduced, should all be proportionate to the goals sought to be achieved and should be time-limited.
Regarding prisons and probation specifically, it’s important to point out that we have adopted Revised European Prison Rules. It is the fourth time they have been revised since 1973, based on the court’s case law and recommendations from the Committee for the Prevention of Torture and Inhuman or Degrading Treatment or Punishment (CPT). This revision considers, in particular, the specific situation of women in detention, foreign prisoners, and the issue of solitary confinement.
We know that these Prison Rules are incorporated as binding in many Member States, although they are what we call soft law because they are not a treaty. Some prison administrations even have a prize for prisons that most fully respect them to encourage prison governors to take them very seriously.
Another significant and very important achievement in the current context is our MEDICRIME Convention against counterfeit medicine, which I think the pandemic has demonstrated to be even more relevant than before. This Convention is vital to protect public health and human rights, to ensure that if you get a vaccine, it’s genuine and approved, or that, if you buy a protective mask, it has been adequately tested, for example. So, the MEDICRIME Convention, which has been increasingly ratified, is something we’ve been putting a lot of emphasis on.
Corruption remains a big issue for my Directorate as well. The work of GRECO-Group of States Against Corruption, which now has 50 parties (the 47 member States of the Council of Europe, plus Belarus, the United States and Kazakhstan), is critical because it is evident that, with the pandemic and its economic consequences, there’ll be a lot of money spent by governments to assist their economies, and sadly, we know that organised crime is always trying to get in there. Corruption is a constant threat. For instance, there’s currently a massive scandal involving politicians who profiteered from masks procurement deals in some Member States.
Therefore, the fight against corruption remains extremely important. In particular, we need to make sure that large sums of public funds are being spent correctly in times of pandemic.
And the third aspect is the continued fight against terrorism because we have seen that terrorist attacks have also continued during the pandemic. Sadly, terrorism remains a very serious threat to our free societies. We need to be constantly aware of and alert to its risks, even if COVID-19 in the news headlines overshadows this issue.
Regarding prisons and probation specifically, it's important to point out that we have adopted Revised European Prison Rules. It is the fourth time they have been revised since 1973, based on the court's case law and recommendations from the Committee for the Prevention of Torture and Inhuman or Degrading Treatment or Punishment (CPT). This revision considers, in particular, the specific situation of women in detention, foreign prisoners, and the issue of solitary confinement.
What are the main concerns and goals of your Directorate regarding the use of artificial intelligence?
JK: Artificial intelligence (AI) is increasingly taking over government functions, with governments delegating decisions, traditionally taken by humans, to artificial intelligence systems. And in the private sphere, companies also roll out AI to a considerable extent. When AI is used to suggest you a song playlist on Spotify, it won’t have a significant impact on your human rights and freedoms, but if your job application with the government or if your case before a court are decided through an algorithm, then all sorts of issues can arise.
We have seen, for instance, clear cases of gender discrimination through the use of algorithms. Just before the pandemic, Amazon announced that their human resources department would no longer be using AI for recruitment – the reason being that their algorithms were based on Amazon’s existing staff structure, which had no women in senior positions. That led the algorithm to conclude that women were not fit for senior positions and automatically excluded all female candidates for senior positions. That is actually and sadly true.
We know from cases in the judicial field, for instance, in the United States, that algorithms based on actual numbers of detainees of a specific background often concluded that it was the Afro-American suspect who should not be released on bail or should be interrogated, and that’s because the predictions that algorithms can make and the conclusions they can draw are based on data, and if data are biased, the outcome will also be biased. So, it’s imperative to ensure that there is a fair process.
As AI is increasingly used, shouldn’t we ensure that the same standards that apply daily to human processes are also applied to automated systems? For example, we demand that judges undergo extensive and thorough judicial training, so we are talking about very high qualifications. Suppose judicial decisions are going to be delegated to machines. In that case, it is better to make sure that the machines, the systems they use and the people who develop and program them also have very high qualifications.
Another sensitive issue related to AI is transparency. When you deal with an algorithm instead of a human being, you must be informed. In this regard, the United States is at the forefront. The State of California already has a law that applies to companies with more than one million customers, which requires them to inform the customer if the interaction is with a bot instead of a human. That must be made clear even more so when State agencies are using AI.
Many of the concerns around using AI are already covered, firstly, in part, by existing legislation, namely the European Convention on Human Rights, which also applies online/to machines.
Second, there is a whole series of so-called Ethical Charters, joint nonbinding statements, from governments, industry and civil society, on the ethical use of AI. But, as noted, they are not binding, so they do not offer the guarantees we need. Scandals like the Facebook-Oxford Analytica scandal demonstrate how important it is to have effective AI and data collection rules.
Therefore, we are now working based on existing non-binding instruments to identify the requirements for a basic treaty on AI – possibly a Convention. We aim to define clear rules for all stages: from design through development to application.
And why the Council of Europe? Although we are a regional organisation, we have set global benchmarks in the field of technology before. At the moment, there is only one treaty in the world on cybercrime: it’s the Council of Europe’s Cybercrime Convention, ratified by sixty-five countries, many outside Europe. In addition, last year we were active with this Convention in more than 130 countries – it is a global standard.
Data protection is another example: forty years ago, the Council of Europe was the first international organisation to establish a data protection convention, which is, I would say, the precursor to the GDPR. Yet another example in the technology sector was the cloning of Dolly, the sheep. Following this event, the Council of Europe quickly negotiated a protocol to ban the cloning of humans. We were able to react very quickly to technological advances.
For all these reasons, the Council of Europe is very well placed to prepare such an instrument regarding the use of AI. We conducted a Feasibility Study that the Member States unanimously approved in December 2020 and is now the basis for the work in progress.
(...) artificial intelligence is already used by some prisons administrations to allocate cells to prisoners (...) artificial intelligence can broadly support – and for the good of all – processes in the context of resocialisation, rehabilitation and programming. In prison systems, AI can have many uses to save time, and it can also save lives. For instance, in prisons where there is a high risk of suicide, AI systems can be used to detect behaviours and signals and, therefore, be an asset to safety.
JT: In September 2019, the Committee of Ministers of the Council of Europe set up the Ad Hoc Committee on Artificial Intelligence (CAHAI).
What is the reason for this special committee, and to what extent does it add value to your Directorate’s work in this field?
JK: My Directorate provides the Secretariat to the CAHAI, a subordinate body of and reports to the Committee of Ministers. Thus, I am responsible for the organisation and management of CAHAI, which comprises senior representatives from government, industry, and civil society, from the 47 Member States and five observer States plus Israel. The latter made a specific request to be admitted to negotiations.
CAHAI was given a mandate at the end of 2019 for two years, and the Feasibility Study was presented halfway through the mandate. Accordingly, we have until the end of 2021 to offer the elements for a Treaty to the Committee of Ministers. And we hope that, during this year, the Committee of Ministers will decide to start formal negotiations.
And then, we will see how quickly the process will unfold. Let me say that the Council of Europe has the record for the fastest treaty ever: after the terrorist attacks in Bataclan and Charlie Hebdo, in seven weeks, we were able to negotiate a binding protocol on foreign terrorist fighters. However, negotiations at the level of a Treaty may take a little longer. We hope that a Treaty can be on the table in the next two and a half years.
Specifically concerning the use of artificial intelligence by the justice systems, including prison and probation services, what are the primary considerations to date, and what is needed to ensure that AI is applied responsibly and in compliance with the standards protected by the Council of Europe, namely human rights, democracy and the rule of law?
JK: It is clear that artificial intelligence (AI) is also considered in the penitentiary field. At the latest Annual Conference of Directors of Prison and Probation Services, we saw an interesting contribution from Singapore, a fully automated prison. That prison does not have guards; AI and robots do everything. I don’t think that’s the model that everyone would want, but artificial intelligence is already used by some prison administrations to allocate cells to prisoners and to decide who should share a cell with whom, who would be compatible, who would pose a risk.
Artificial intelligence can broadly support – and for the good of all – processes in the context of re-socialisation, rehabilitation and programming. In prison systems, AI can have many uses to save time, and it can also save lives. For instance, in prisons where there is a high suicide risk, AI systems can be used to detect behaviours and signals and, therefore, be an asset to safety. But, as I said before, data are crucial, and we must ensure that AI does not lead to discrimination or restrict the rights of detainees.
JT: At the 25th Council of Europe Conference of Directors of Prison and Probation Services (CDPPS), held virtually on 10-11 November 2020, you mentioned that correctional administrations should invest in adequate compensatory measures to alleviate the restrictions on visits, contacts, interventions, and activities caused by the pandemic crisis.
To what extent do you consider that respect for and protection of offenders’ human rights were affected across the CoE Member States because of the COVID-19 pandemic? How does your Directorate support member jurisdictions to meet the upheld standards despite the sanitary crisis?
JK: It is clear that many people living together in limited spaces and closed environments, such as prisons, are subject to greater risk, and therefore, since the beginning of the pandemic, we have issued guidelines to help national authorities deal with the crisis. In addition, CPT has continued to visit prisons to monitor the situation and ensure that appropriate measures are being taken regarding COVID-19.
We also have a bioethics committee in the Council of Europe, which drafted guidelines on ethical vaccination. This committee emphasised that particularly vulnerable groups – and people in detention certainly are vulnerable – should be duly given priority in vaccination.
There are several fronts on which we have been active. Moreover, we are currently considering guidance to our Member States on the mental health of prisoners, which we hope to issue soon. Although the pandemic is tough for everyone, it is particularly harsh for detainees, mainly because they tend not to have easy access to digital communication tools and cannot see their families in person during this period.
So, we are dealing with many issues that are very relevant in times of pandemic and major challenges both posed by the pandemic and by the reaction to it.
Director of the Information Society and Action against Crime Directorate, Council of Europe
Jan Kleijssen joined the Council of Europe in 1983 as a lawyer with the European Commission of Human Rights. In 1987, he was appointed to the Secretariat of the Parliamentary Assembly. Kleijssen was Secretary to the Political Affairs Committee from 1990 to 1999. He then served as the Director of the Secretary General’s Private Office until 2004. Subsequently, Kleijssen served as Director in the Parliamentary Assembly and Special Advisor to the President. In 2006, he moved to the Directorate General of Human Rights and was Director of Standard-setting until 2011, when he was appointed to his current position. He holds a degree in International Law from Utrecht University and a MA in International Affairs from Carleton University.