Interview with Dr. Karine Caunes at CPDP.ai 2024
This interview was held before CPDP.ai 2024 and was originally published by CPDP in the lead-up to the conference. Hiba Harchaoui for CPDP: Could you please tell us a bit about yourself?
Karine Caunes: At the Center for AI and Digital Policy (CAIDP), I currently serve as Global Program Director. Together with Marc Rotenberg and Merve Hickok, I am teaching in our policy clinics. I am also in charge of CAIDP AI & Democratic Values (AIDV) Index, evaluating the public policies of 80 countries and I have been representing CAIDP in various settings such as UNESCO expert group meetings on AI or the plenary sessions of the Council of Europe Committee on AI for the negotiation of the Convention on AI, human rights, democracy and the rule of law.
Regarding CAIDP Europe, I’m assuming the role of Executive Director. As a European law scholar, cur- rently Editor-in-Chief of the European Law Journal – which recently released a special issue on “Law and the Common Good in the Digital Age” –, developing CAIDP’s European branch is a task I feel passionate about and I am grateful to be collaborating with a very talented team on this project. I will spearhead CAIDP Europe’s research and policy work. The establishment of CAIDP Europe stems from the growing number of activities we have undertaken in Europe and the rapidly evolving policy and legal landscape in the region. We believe that incorporating CAIDP into a dedicated branch is the most effective way to advance our mission of promoting human-centric AI regulation now that Europe is switching gears from policy-making to implementation.
Can you tell us a bit more about the Center for AI and Digital Policy (CAIDP)? Its goals and its rapid global spread?
Karine Caunes: The Center for AI and Digital Policy (CAIDP) aims to promote a better society that is more fair and more just; a world where technology fosters broad social inclusion, based on fundamental rights, democratic institutions, and the rule of law. CAIDP evaluates national AI policies and practices, trains AI policy leaders, and advocates for democratic values in AI governance. In terms of growth, we can categorize it into four main pillars: education, research, advisory, and advocacy. Regarding education, our first policy clinic in Spring 2021 had 11 participants from three countries. Fast forward to Spring 2024, we just graduated our Spring cohort with nearly 300 participants from 80 countries. This includes the expansion of our policy clinics which now encompasses an advanced level as well as a policy and legal group. Our participants come from diverse backgrounds, including law, policy, technology, journalism, or civil society, reflecting the multidisciplinary approach we advocate for. Our aim is to train the next generation of AI policy leaders who can make a meaningful impact in their respective fields and countries, all while prioritizing a human-centric perspective.
In terms of research, we publish the AIDV Index annually. This report evaluates the development of AI policy around the world based on key metrics among which the endorsement and implementation of the most influential AI policy frameworks, namely the OECD AI principles and the UNESCO Recommendation on the ethics of AI. This international perspective is crucial as the challenges posed by AI are inherently global and we need to foster upward convergence towards a human-centric governance of AI. Understanding international frameworks also helps shape national policies.
Related to this, we have developed an advisory role towards national governments and international organisations such as the UN, including UNESCO, the G7, the G20, the OECD, and the EU. In this role, we assist these organisations in creating AI policies that align with democratic values and help to set global standards. With regard to advocacy, if I take the example of the Council of Europe Convention on AI, we have led a global campaign with civil society organisations and academics to call on the EU and negotiationg states, both from within and outside Europe such as the US, Canada, Japan, and various Latin American countries, to ensure that the Convention applies equally to the public and the private sectors. From my perspective, the added value of this convention lies not only in being the first-ever binding treaty with a human-centric approach but also in establishing an institutionalised mechanism of cooperation among state parties. From an EU perspective, another significant aspect is that the Convention can serve as an interpretative tool regarding EU AI-related legislation. This might be key in ensuring an overall rights-based approach, akin to that enshrined in the GDPR, and addressing some legislative loopholes. However, what we need to be vigilant about, is that European standards are not being lowered under the cover of international “interoperability”. Partnerships with like-minded organisations and individuals across various sectors worldwide further amplify our advocacy efforts, with the aim to rebalance the power dynamics between tech giants and citizens or even democracies.
In the same spirit, we also promote public voice opportunities. Public participation is key. As a civil society organisation and with other civil society organisations, we are monitoring and trying to contribute in the way AI governance is currently being shaped. I would say that 2023 was particularly pivotal, especially with the spread of OpenAI ChatGPT, and the likes, which sparked significant concerns and policy reactions. Already in March 2023, CAIDP filed a complaint to the US Federal Trade Commission (FTC) regarding bias, transparency, privacy, safety, and deception risks of Open AI ChatGPT as well as lack of independent oversight and has called on the FTC to issue an order to establish guardrails for ChatGPT. We’ve seen a slew of investigations by data protection authorities in Italy, Canada, France, Australia, Germany, Spain and Colombia. The G7 launched the Hiroshima process addressing generative AI. However, beyond the hype focusing on generative AI, there has been a broader positive policy impact: states shifted from questioning whether to regulate AI to concentrating on how to regulate it. For CAIDP and CAIDP Europe, this signifies a transition from policy-making to the implementation and enforcement phases. Now that regulations are being put into place, we need to operationalise them effectively.
Overall, all our activities involve not only identifying issues and highlighting best practices but also promoting dialogue and collaboration. By fostering a culture of mutual learning, we aim to contribute to shaping AI governance in a manner that upholds human-centric values and principles across borders.
Could you provide more details about CAIDP Europe, particularly the opening of the European branch?
Karine Caunes: Our focus on ensuring human-centric governance of AI translates into upholding a rights-based approach to AI regulation in Europe. The primary goal of CAIDP Europe is to further the mission of CAIDP at the level of the Council of Europe, the European Union, and European states. This involves engaging with policymakers, supporting the implementation of AI-related laws, and collaborating with civil society organisations, academics, data protection authorities, other regulatory agencies set up to enforce the new EU AI legislative package, and national human rights institutions. In short, the upcoming adoption of the Council of Europe Convention on AI and the EU’s adoption of an AI legislative package has emphasised the need for us to strengthen our presence in Europe in order to carry on impactful policy actions.
A key focus for CAIDP Europe will be on the EU AI Act. This legislation has a defined timeline for its applicability, and it’s crucial for us to closely monitor its respect. One important aspect is the designation of national surveillance market authorities. These authorities will play a significant role in enforcing the regulation. We’ll collaborate with them, data protection authorities, national human rights institutions, and civil society organisations to ensure effective oversight. In assessing the role of these authorities however, independence is a crucial parameter. For instance, the EU AI Office has been created within the European Commission and the structure of most future national supervisory authorities is still unknown. The Netherlands provided a good practice with the creation of a dedicated Department for the Coordination of Algorithmic Oversight within its independent Data Protection Authority. In France, the CNIL also created a special unit but it is still unclear whether it will be the competent authority with regard to the EU AI Act. This is without mentioning the enforcement system set up by the Digital Services Act (DSA) or the GDPR which all will have to be coordinated. Our AIDV index covers 24 European countries, providing a comparative view of AI governance. CAIDP Europe will closely examine and asses the governance landscape in these countries, considering factors beyond just the titles of designated authorities.
The launch of CAIDP Europe will take place at the CPDP.ai conference in May. During the opening night on May 21st, we’ll present AI Policy Leader Awards to recognise exemplary efforts to safeguard human rights and uphold democracy and the rule of law in the age of AI. We’ll also host a workshop on May 23rd to strategise on the implementation and enforcement of human-centric AI. This collaborative approach, involving civil society organisations, academics, data protection authorities, human rights bodies and likeminded partners, aims to create synergies and maximise impact. Everybody is also welcome to join us for a cocktail we will be sponsoring in that evening to further exchange views and discuss possible collaborations. We have now launched the CAIDP Europe website and are establishing our presence on LinkedIn to facilitate ongoing engagement. CAIDP Europe’s incorporation as a nonprofit association in Brussels symbolises our commitment to serving the entire European region, including EU and non-EU countries. This broader perspective acknowledges the regional and geopolitical dimensions of AI governance. In summary, CAIDP Europe’s launch will occur during the CPDP conference in May, with a focus on fostering collaboration and strategising for human-centric AI governance across Europe.
Could you outline five long-term goals for CAIDP Europe?
Karine Caunes: Our primary long-term goal is to ensure human-centric AI governance in Europe and promote it across all levels, including Europe- an, national, and local. We aim to create synergies with existing ecosystems that share our vision of fostering a rights-based approach to AI regulation in Europe. This involves working with civil society organizations, academics and other partners to flag issues, find concrete solutions and provide support where needed. Our approach is collaborative and bottom-up, recognising the importance of considering local contexts and fostering convergence for impactful policy actions. By fostering dialogue and cooperation, we can work towards common goals and address challenges effectively.
Adopting a rights-based agenda in Europe is key to tackle the challenges posed by AI for several reasons. First, it is part of Europe’s DNA. After the Second World War, European integration has developed through law and through the conferral of rights to citizens in order to defend our most fundamental values. Second, since at least the Enlightenment, innovation has been defined and measured by reference to the betterment of the human condition and to its contribution to the common good. Thus, opposing innovation to human rights protection, as a currently widespread narrative would have it, is a contradiction in terms. It is the negation of the humanistic spirit that characterises European culture. Learning from the dark side of our history would mean to stop repeating ad nauseum this presumption of an inevitable contradiction between innovation and human rights protection which assumes at the same time that we have to accept risks, and in a slippery slope human rights violations, and which in effect limits accountability. Advocating for a rights-based approach to AI regulation is to shed light on this pervert logic and put back the human at the start, center and end of innovation.
To define our strategy, we have mapped out 3 types of challenges the AI revolution poses to the rule law, democracy and fundamental rights: the privatisation of power; the authoritarian temptation; and the lack of proper enforcement. Preventing them through a pro-active approach is our main objective. We have selected our policy priorities accordingly: Ensuring human centric innovation; allowing fair and inclusive democracies to thrive; and consolidating a human-centric governance of AI. The latter involves bringing clarity in the European governance maze and mapping who is responsible for what in terms of enforcement to ensure the effective protection of fundamental rights. As for democracy and the rule of law, of particular concern is the development of multi-use AI. The large carve out in the EU AI Act regarding the use of AI for national security purposes, despite the risks mass surveillance practices pose to our freedoms, denotes States’ outdated vision of minimising their obligations. On the contrary, setting up safeguards vital to our democracies is a right States have against foreign influence and a right citizens have as the source of sovereignty in democratic states. Talking about foreign influ - ence and preserving democracy, the DSA is already being put to the test with various elections taking place in Europe in 2024. Ensuring the DSA plays its role will be one of our objectives. But multi-use AI also concerns the digitalisation of public administrations. Public authorities should be mindful of setting up the necessary safeguards regarding the use of commercial AI to deliver public services. Decisions taken by public authorities do have a direct impact on citizens’ lives and their trust in public institutions. Some grim examples include the Dutch childcare benefits scandal or the UK Post Office scandal. Promoting best practices such as the adoption of AI registers by European municipalities and extending and transposing this experience to public administrations will also be one of our priorities.
But beyond multi-use AI, it needs to be acknowledged that AI systems are created for the most part by private companies. Ensuring human-centric innovation is thus fundamental. CAIDP Europe will advocate for some key principles to be operationalised such as human-centric AI by design. This includes the protection throughout Europe of the right to algorithmic transparency; data protection rights, quality and sound governance; the right to non-discrimination; as well as the implementation of fundamental rights impact assessments whether under the EU AI Act or the DSA. Structural inequalities, especially those which come from power relationships or society, such as those regarding the rights of workers, marginalised communities, people with disabilities, or children, also need to be taken into consideration. CAIDP Europe will focus on human-centric accountability and will put to the test citizens’ right to be informed about the use of AI; the right to an effective remedy through individual and collective action and the right to a fair trial. CAIDP Europe will monitor closely CJEU and ECtHR cases and engage in strategic litigation.
CAIDP Europe will also advocate for European and State authorities to adopt the necessary measures to ensure the right to a sustainable environment towards a successful twin green and digital transition. The recent case law of the European Court of Human Rights in this regard is particularly encouraging.
In a nutshell, we recognise the limitations of a solely risk-based approach and seek to contribute to the implementation of a robust rights-based approach to AI regulation in Europe. A human-centric interpretation and application of the EU AI Act, in sync with data protection and other AI-related legislation, will be key. We will be mapping issues, possible solutions and best practices under the EU Charter of Fundamental Rights and the Council of Europe European Convention on Human Rights. This will guide our strategy and actions. The AI revolution should be an opportunity to better uphold our fundamental values. The stakes are high and the challenge immense but CAIDP Europe is com- mitted to playing its part in this revolution togeth- er with like-minded partners. We look forward to meeting them at the CPDP.ai conference and further defining our collective contribution towards a human-centric governance of AI in Europe.
