The Council of Europe announced the AI Convention yesterday. This treaty supposedly promotes responsible innovation while addressing the myriad risks AI might pose.

The AI Convention is primarily concerned with safeguarding the human rights of those affected by AI systems. It’s a separate entity from the EU AI Act, which came into force last month and imposes stringent regulations on AI development, deployment, and use within the EU market.

The Council of Europe, established in 1949 and distinct from the EU, boasts 46 member countries, including all 27 EU states. In 2019, an ad hoc committee began exploring the feasibility of an AI framework convention, and by 2022, a Committee on Artificial Intelligence had drafted and negotiated the text.

Signatories can adopt or maintain legislative, administrative, or other measures to implement the provisions.

Francesca Fanucci, a legal expert at the European Centre for Not-for-Profit Law Stichting (ECNL) who helped draft the treaty alongside other civil society groups, told Reuters the agreement had been "watered down" into a vague set of principles.

"The formulation of principles and obligations in this convention is so overbroad and fraught with caveats that it raises serious questions about their legal certainty and effective enforceability," she said.

 Fanucci pointed out significant flaws, such as exemptions for AI systems used for national security purposes and the limited scrutiny of private companies compared to the public sector.

"This double standard is disappointing," she added.