⚡ Quick Summary
Published by the Council of Europe in its Treaty Series as No. 225, this explanatory report accompanies the Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law, adopted by the Committee of Ministers on 17 May 2024 and dated Vilnius, 5 September 2024. It comments on the Preamble and on Articles 1 to 36 and states that it does not constitute an instrument providing an authoritative interpretation of the Convention.
The report explains that the Convention's object is to ensure that activities within the lifecycle of artificial intelligence systems are fully consistent with human rights, democracy and the rule of law, applying each Party's existing obligations rather than creating new human rights. It records the OECD definition of an AI system of 8 November 2023, the "adopt or maintain" and "graduated and differentiated" approach to measures, the illustrative lifecycle activities from planning and design to retirement, and the exemptions for national security, national defence and certain research and development.
The commentary covers general obligations, seven principles including human dignity and individual autonomy, transparency and oversight, accountability, equality and non-discrimination, privacy, reliability and safe innovation, remedies and procedural safeguards, risk and impact management with possible moratoria or bans, implementation measures, and the follow-up mechanism with a Conference of the Parties, reporting and international co-operation.
🧩 What’s Covered
The report follows the Convention's structure, commenting on the Preamble and each article in turn.
- Introduction and Preamble: the mandate to the Committee on Artificial Intelligence after the CAHAI's 2021 paper, the 68 civil society and industry observers, and the instruments taken into account, including the OECD AI Principles, CM/Rec(2020)1, the UNESCO Recommendation on the Ethics of AI, the G7 Hiroshima Process and the EU AI Act.
- Chapter I – General provisions (Articles 1–3): object and purpose limited to the Council of Europe's mandate; the eight illustrative lifecycle activities from planning and design to retirement; the "adopt or maintain" and "graduated and differentiated" approach; the AI system definition drawn from the OECD revision of 8 November 2023; and scope covering public authorities, private actors, Party declarations and the national security, national defence and research exemptions.
- Chapter II – General obligations (Articles 4–5): the duty to respect, protect and fulfil human rights, with lists of United Nations, Council of Europe and EU, and Inter-American instruments, and obligations on democratic processes and the rule of law, naming risks to separation of powers, political pluralism, elections and judicial independence.
- Chapter III – Principles (Articles 6–13): human dignity and individual autonomy; transparency and oversight, including explainability, interpretability, labelling and watermarking; accountability and responsibility; equality and non-discrimination with five documented sources of bias; privacy and personal data protection; reliability through standards and assurance schemes; and safe innovation through regulatory sandboxes.
- Chapters IV and V – Remedies and risk management (Articles 14–16): effective and accessible remedies and procedural safeguards; threshold tests such as "significantly affect human rights" and "substantially informed by the use of the system"; notification of interaction with an AI system; and an iterative risk and impact management framework with documentation, testing and moratoria or bans.
- Chapters VI and VII – Implementation and follow-up (Articles 17–26): non-discrimination in implementation, rights of persons with disabilities and children, public consultation, digital literacy and skills, safeguards for existing human rights, wider protection, the Conference of the Parties, reporting within two years of becoming a Party, and international co-operation.
- Chapter VIII – Final clauses (Articles 27–36): effects of the Convention including EU law, amendment, dispute settlement, signature and entry into force after five ratifications, accession, territorial application, the federal clause, reservations, denunciation and notification.
💡 Why it matters?
For states, regulators, compliance teams and oversight bodies, the report explains how a binding Council of Europe instrument translates existing human rights, democracy and rule of law obligations into the AI lifecycle. It gives the reasoning behind duties to assess and mitigate risks, to document and explain system behaviour, to keep remedies and procedural safeguards effective, and to designate functionally independent oversight. The commentary connects the Convention to named regimes — the ECHR, the ICCPR, Convention 108+, the GDPR and the EU Artificial Intelligence Act — and shows the flexibility Parties retain through the "adopt or maintain" and graduated and differentiated approach, which matters when mapping national implementation choices.
❓ What’s Missing
The report is a commentary, not the treaty text: the Convention's articles are paraphrased rather than reproduced, so readers need the Convention itself. It leaves decisive terms open — what counts as a "significant effect" on human rights, what use is "incompatible" enough to warrant a moratorium or ban, and how declarations on private-sector coverage will work in practice. Practical detail on enforcement, penalties and the format of Party reports is deferred to the Conference of the Parties. The text dates from 5 September 2024 and records neither ratifications nor entry into force.
👥 Best For
Best for legal and policy advisers mapping how the Convention's obligations apply to AI activities, for compliance and risk teams building impact assessment, documentation and transparency processes across the lifecycle, and for oversight bodies, national human rights institutions and regulators preparing supervisory roles and the public consultation, digital literacy and reporting duties.
📄 Source Details
Explanatory Report to the Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law, Council of Europe Treaty Series No. 225, dated Vilnius, 5.IX.2024; published by the Council of Europe, with no individual authors named. The report runs 33 pages in English and accompanies the Framework Convention adopted by the Committee of Ministers on 17 May 2024. The input was the full 33-page text extraction. No URL for the document is printed in the text.
