In September 2024, the Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law, known as the AI Convention or AI Treaty, was introduced for signature by the Council of Europe (CoE). This landmark treaty aims to govern the ethical development and use of artificial intelligence on an international scale, following the earlier enactment of the European Union's AI Act. The AI Convention stands out as the first international framework of its kind, seeking to influence AI regulation beyond the confines of the European region.
The EU AI Act and the AI Convention share several foundational principles, yet they diverge in some key areas. While the EU AI Act provides detailed regulations on General Purpose AI (GPAI) systems and is focused on the specific obligations of products within EU member states, the AI Convention adopts a broader perspective. It prioritises the preservation of human rights, democracy, and the rule of law without directly addressing GPAI systems. The EU AI Act is prescriptive and includes enforceable regulations with applicable penalties, whereas the AI Convention offers broad commitments leaving specific compliance to be determined by national legislation.
The AI Convention is designed to encourage wider international adoption and cooperation around AI regulation—it is open to CoE members and non-members alike, thus having a potentially global reach. It employs a risk-based approach that assesses AI impacts specifically in relation to human rights, democracy, and the rule of law, rather than categorizing operational risks explicitly as seen in the EU AI Act.
Several countries, including the United States, Canada, Japan, Australia, and the United Kingdom, have already signed the AI Convention. These signatures indicate a commitment to a range of ethical imperatives, including the prioritisation of human rights, non-discrimination, data protection, transparency, accountability, and oversight of AI systems. Additionally, the treaty supports innovation by suggesting the use of regulatory sandboxes and allowing certain flexibilities for private sectors, which can either follow the AI Convention obligations or adopt suitable alternative measures. However, these privileges are balanced by strict adherence to human rights standards.
Despite the promising outlook, the AI Convention cannot enter into force without fulfilling certain prerequisites: it mandates at least three signatures from CoE member states and five ratifications by signatories, a process involving domestic legislative assent. After fulfilling these requirements, a three-month period is necessary before the treaty becomes effective.
Some criticisms have been directed at the AI Convention, particularly regarding its enforceability and possible loopholes. Francesca Fanucci, a legal expert at the European Center for Not-for-Profit Law who participated in drafting the treaty, cautions that the Convention's broad principles might prove challenging to implement and enforce uniformly. She points to exemptions related to national security and the limited oversight of private enterprises as potential areas where loopholes could undermine the framework.
The CoE Secretary General, Marija Pejčinović Burić, has advocated for more countries to join and for early signatories to pursue ratification swiftly, which could pave the way for the Convention's entry into force. As the AI landscape continues to evolve rapidly, the AI Convention marks a significant step toward fostering ethical practices and international collaboration in AI development and utilisation. Further steps are anticipated in addressing concerns and refining the treaty for broader global adherence and practical enforcement.
Source: Noah Wire Services