On 11 March 2026, the European Parliament approved the EU's conclusion of the Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law. The Convention, known by its treaty designation CETS 225, is the first legally binding international treaty governing the development and deployment of artificial intelligence.
It was opened for signature on 5 September 2024. The EU's formal conclusion of the Convention follows the Commission's earlier signing and now brings the EU within the treaty's binding obligations.
What the Convention Establishes
The Convention's stated aim is to ensure that every stage of an AI system's lifecycle is fully consistent with human rights, democracy, and the rule of law. It applies to both public authorities and, critically, private actors operating within its parties' jurisdictions.
The principles codified in the treaty cover accountability, requiring that mechanisms exist to hold AI developers and deployers accountable for harms. Non-discrimination, requiring that AI systems must not discriminate on protected grounds. Privacy and data protection, requiring that AI lifecycle activities comply with existing privacy frameworks. Reliability and safety, requiring that AI systems must be designed and operated to perform as intended without causing harm. Transparency, requiring that individuals must be able to understand when AI is affecting decisions that concern them. Procedural safeguards, requiring that individuals must have access to remedies where AI activity affects their rights. And democratic oversight, requiring that parties maintain the capacity for democratic and judicial review of AI deployments.
The treaty requires parties to define in domestic law the purposes for which AI may be used, the categories of data that can be processed, and the safeguards available to individuals. Independent oversight is mandated. The Convention explicitly states that compliance cannot rest on administrative discretion: it requires transparent rules and enforceable guarantees.
Scope and Entry Into Force
The treaty will enter into force on the first day of the month following the expiration of three months after five signatories, including at least three Council of Europe member states, have ratified it. The EU's parliamentary approval on 11 March 2026 is a significant step towards this threshold.
The Convention is notable for its international scope: it is open to non-European countries to accede, meaning its standards may become a reference point for AI governance well beyond the EU.
HumanSafe Opinion
The Convention's requirement for "transparent rules and enforceable guarantees" is the most important phrase in the treaty for understanding what rights-aligned AI actually requires. It establishes that governance depending on administrative discretion is not sufficient, that the standard must be structural, transparent, and legally enforceable, not dependent on any particular operator's goodwill or any particular regulator's willingness to act.
This is a constitutional standard expressed in international law. It is the recognition, at the highest level of international legal authority, that rights-based constraints on AI must be architectural properties rather than policy commitments. When the first legally binding international AI treaty describes the future of responsible AI in constitutional terms, it is not describing an aspiration. It is describing a requirement. The question for every organisation building or deploying AI is whether their systems meet that standard, and by what mechanism the guarantee is actually enforceable.
Sources
- Texts adopted: Council of Europe Framework Convention on AI and Human Rights — 11 March 2026 — European Parliament, March 2026
- The Framework Convention on Artificial Intelligence — Council of Europe
- Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law (CETS 225) — Council of Europe full treaty text
- The Framework Convention on Artificial Intelligence: Embedding Human Rights in the Digital Age — Federal Bar Association, 2025
