Council of Europe approves AI Convention, but not many reasons to celebrate

The world’s first AI treaty was approved in May by the Council of Europe (CoE). It sought to protect human rights, democracy, and the rule of law. However, the final result has fallen short of the goals.

Despite repeated calls from civil society, including EDRi-member European Center for Not-for-Profit Law (ECNL), for comprehensive human rights protections, the Convention comes with significant flaws.

What are the Convention’s main shortcomings?

It allows States to exclude all AI systems designed, developed or deployed for the protection of national security interests from its application. Furthermore, the Convention fails to include the preservation of health and the environment as one of the general principles underpinning the activities within the lifecycle of AI systems. Overall, the Convention introduces imprecise language on States’ obligations that raises serious questions about their legal certainty and concerns about their enforceability and effectiveness.

These flaws and omissions could have been significantly improved. The Parliamentary Assembly of the Council of Europe (PACE) Committee proposed constructive amendments to the text, which were echoed by the UN High Commissioner for Human Rights and the Special Rapporteur on Environmental Defenders under the Aarhus Convention, respectively.

What are the next steps?

The Convention will be opened for signature on 5 September 2024. The CoE will develop a methodology for the impact assessment of AI systems on human rights, democracy and rule of law. ECNL will continue monitoring the follow-up work of the CoE and states, and invoke more meaningful cooperation with the other stakeholders, including civil society, to secure its strongest possible implementation.

Contribution by: EDRi member, ECNL