The European Union has signed the first legally binding international treaty on Artificial Intelligence. This is a convention promoted by the Council of Europe, to which the United States and the United Kingdom, among others, have also adhered. The Convention provides a common approach to ensure that AI systems are compatible with human rights, democracy and the rule of law, while enabling innovation and trust by citizens.
The signatories
The text includes a series of key concepts also present in European legislation on AI, such as the risk-based approach, transparency of the content generated by this technology, detailed documentation obligations for systems identified as high-risk with the possibility of introducing bans for those considered a clear threat to fundamental rights.
In addition to the EU, the US and the UK, Canada, Mexico, Japan, Israel, Australia, Argentina, Peru, Uruguay, Costa Rica and the Vatican have also signed up. “This Convention is an important step towards ensuring that these new technologies can be exploited without eroding our most ancient values, such as human rights and the rule of law,” said the British Justice Secretary, Shabana Mahmood, in a statement.
Final green light for the world’s first AI law: what changes
The AI Convention focuses primarily on protecting the human rights of people affected by AI systems and is separate from the EU directive, the AI Act, which came into force last month. The Council of Europe, founded in 1949, is an international organization separate from the European Union with a mandate to safeguard human rights: 46 countries, including all 27 EU member states, are members.
Writing the text
In 2019, an ad hoc committee began examining the feasibility of a framework convention on artificial intelligence, and in 2022, an Artificial Intelligence Committee was established to draft and negotiate the text. Signatories may choose to adopt or maintain legislative, administrative or other measures to implement the provisions.
Francesca Fanucci, a legal expert at the European Center for Not-for-Profit Law Stichting (ECNL), who helped draft the treaty along with other civil society groups, said the agreement had been “watered down” into a broad set of principles. “The wording of the principles and obligations in this convention is so broad and full of caveats that it raises serious questions about their legal certainty and actual enforceability,” Fanucci told Reuters, highlighting the exemptions for AI systems used for national security purposes and the limited oversight of private companies compared to the public sector as flaws.
