Council of Europe AI Convention Enters Force
Summary
On November 1, 2025, the Council of Europe's Framework Convention on Artificial Intelligence (CETS 225) entered into force following ratification by the United Kingdom, France, and Norway — the first binding international treaty on AI. The Convention covers the entire AI lifecycle, applies to both public and private actors, and is open to non-Council of Europe states including the United States, Canada, Japan, and Israel. It establishes human rights, democracy, and rule of law as the governing principles against which AI systems must be assessed.
What Happened
CETS 225 was opened for signature in September 2024 in Vilnius, where it was signed by 46 of the 47 Council of Europe member states, plus the US, Canada, Japan, Israel, and Australia. The Convention required five ratifications to enter into force. The UK, France, and Norway deposited instruments of ratification in October 2025, triggering the November 1 entry into force.
The Convention's substantive obligations require parties to establish legal frameworks ensuring AI systems are developed, deployed, and used in ways consistent with human rights, democratic principles, and the rule of law. Specific obligations include: prohibiting or restricting AI applications that present unacceptable risks; establishing oversight mechanisms including judicial and administrative remedies; ensuring transparency and explainability where rights may be affected; and maintaining human oversight and control over AI systems operating in high-stakes contexts.
For EU member states, the Convention operates alongside the EU AI Act, which is more specific and operationally detailed. The two instruments were designed to be compatible: the Convention sets the human rights floor, while the AI Act provides technical and procedural implementation.
The inclusion of non-CoE states — particularly the US — as signatories was significant. While the US had not yet ratified, its signature committed it to good-faith engagement with the Convention's principles and created a reference point for US courts and regulators. US ratification would require Senate advice and consent, which was considered unlikely under the Trump administration.
The European Parliament's recommendation A10-0007/2026, adopted in early 2026, encouraged all EU institutions to ensure EU instruments remained consistent with CETS 225 standards.
Why It Matters
CETS 225 represented a qualitative shift in the international AI governance landscape: from soft law (declarations, voluntary commitments, codes of practice) to hard law with state-level binding obligations. The Convention did not establish an international enforcement body — implementation was left to national legal systems — but it created treaty obligations that states could be held to account for under international law.
The Convention's open membership model was its most strategically significant design feature. By allowing non-CoE states to ratify, it created a potential pathway for a genuinely global binding AI governance treaty that did not require UN-level consensus. Whether the US, Japan, or Canada would ultimately ratify depended on domestic political conditions, but their signatures maintained the possibility.
The human rights framing — rather than the risk-based or sectoral framing used in the EU AI Act or US sector-specific rules — offered a different organizing principle for AI governance. It connected AI regulation to existing human rights legal infrastructure, including monitoring bodies, reporting obligations, and enforcement mechanisms that states had already committed to. This made the Convention potentially more enforceable than instruments without established compliance ecosystems.
For the governance competition narrative, the Convention complicated the two-axis framing (US acceleration vs. EU regulation). A third axis — international human rights treaty law — was now in force and applicable to all ratifying states regardless of their domestic AI policy preferences.