AI Policy Radar

    CoE AI Convention

    The only instrument that could bind the US, EU and UK to common AI obligations. Its flexible 'graduated obligations' for private actors will test whether treaty-based AI governance has real teeth or remains declaratory.

    What it is

    The first legally binding international AI treaty, requiring parties to ensure AI activities respect human rights, democracy and rule of law across the AI lifecycle. It has been signed by the EU, US, UK, Japan, Canada, Norway, Iceland, Israel and others; the EU is so far the only ratifying Party.

    Opened for signature Sep 2024. The EU ratified on 15 May 2026, the first Party to do so. Entry into force requires five ratifications including three Council of Europe member states, which has not yet happened.

    Who is affected

    The only instrument that could bind the US, EU and UK to common AI obligations. Its flexible 'graduated obligations' for private actors will test whether treaty-based AI governance has real teeth or remains declaratory.

    Key obligations

    • Parties must ensure AI lifecycle activities comply with human-rights obligations
    • Risk and impact assessment frameworks; remedies for AI-caused harms
    • Applies to public authorities; parties choose modalities for private actors

    Recent signals

    • 15 May 2026

      EU ratifies the Council of Europe AI Convention

      The EU became the first Party to ratify the Framework Convention, at the Committee of Ministers' session in Chișinău. The treaty still needs five ratifications, including three Council of Europe member states, to enter into force.

    • 5 September 2024

      Council of Europe AI treaty opens for signature; US, EU, UK sign

      The first binding international AI treaty begins gathering signatures and ratifications.

    Related instruments

    ← Back to the AI Policy Radar