On October 13th, 2022, the European Data Protection Supervisor (EDPS) published an Opinion entitled “Recommendation for a Council Decision authorising the opening of negotiations on behalf of the European Union for a Council of Europe convention on artificial intelligence, human rights, democracy and the rule of law”. This independent supervisory authority welcomes the initiative taken by the European Commission to authorise negotiations on behalf of the EU regarding the future Council of Europe’s (CoE) Convention on Artificial Intelligence (AI). The Opinion also provides a series of recommendations on the Commission’s negotiating directives.
The Council of Europe is currently working on a binding legal framework on artificial intelligence based on its standards on human rights, democracy, and the rule of law. Besides the 46 Member States of the CoE, the legal instrument will also be available to non-Member States. This means that the Convention on AI may be the first legally binding international instrument to regulate AI. And, as the EDPS notes, the Convention on AI would have a ‘horizontal nature’ like the AI Act, which means that it will apply to users and providers from both the private and public sectors.
Furthermore, the EDPS acknowledges the EU Commission’s interest in ensuring compatibility between the AI Act and the Convention on AI. Additionally, the independent authority notes that “the convention represents an important opportunity to complement the proposed AI Act by strengthening the protection of fundamental rights of all persons affected by AI systems”. Accordingly, in an effort to ‘reinforce’ the rights of individuals impacted by AI systems, the EDPS suggests the Commission include “a methodology for assessing the risks posed by AI systems on fundamental rights”.
Moreover, the EDPS recommends that special attention be given to AI systems used for criminal justice and law enforcement. To this end, the Opinion recommends that the Commission finds a balance between the public interest and the rights of those subject to AI systems, in order to respect fundamental rights.
The EDPS also recommends that the risk-based approach (which the AI Act also established) considers “societal/group risks posed by AI systems”. Furthermore, the Opinion recommends that the Commission includes in the negotiating directives the prohibition of AI systems that pose an “unacceptable risk” such as those that use social scoring, biometric identification of individuals in publicly accessible spaces, biometric and emotion categorisation, as well as risk assessment systems used by law enforcement, and AI systems with questionable scientific validity.
Apart from these recommendations, the opinion puts forward several other recommendations ranging from the degree of power that should be given to supervisory authorities, to the technical standards that should be applied. However, the Commission will not have the power to negotiate on behalf of EU Member States regarding national security.
It remains to be seen how the two legal frameworks will complement each other in terms of protecting individuals from high-risk AI systems.