Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law (CETS No. 225)

Overview

The Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law (CETS No. 225) is a multilateral, legally binding treaty that establishes obligations for States concerning the design, development, deployment, operation and decommissioning of artificial intelligence systems so as to protect human rights, democratic institutions and the rule of law. The Convention was adopted by the Committee of Ministers of the Council of Europe on 2024-05-17 and opened for signature on 2024-09-05. Uruguay signed the Framework Convention on 2025-09-02 in Strasbourg; the signature was performed by Ambassador Enrique Emilio Loedel Soca following internal executive acts designating him as plenipotentiary. Signature by Uruguay is an important diplomatic step but does not by itself constitute deposit of an instrument of ratification or entry into force for Uruguay under the Convention’s procedures.

Definitions

The Convention uses technology-neutral definitions to determine its scope and obligations. Key definitions used in the Convention and for Uruguay’s application of it include:

Artificial Intelligence System: a broadly defined, OECD-derived concept of a machine-based system that, for explicit or implicit objectives, infers from input how to generate outputs (predictions, content, recommendations or decisions) that may influence physical or virtual environments; systems vary in autonomy and adaptiveness.
Lifecycle: all stages of an AI system from design, research and development, testing, deployment, operation, maintenance and monitoring through to decommissioning.
Party/State: a State that has signed and deposited its instrument of ratification, acceptance, approval or accession to the Convention and thereby expressed consent to be bound, under the procedures established by the Convention.

Governance and Institutional Framework

The Convention establishes a governance architecture built around State-level implementation and intergovernmental follow-up. Parties are required to adopt legislative, administrative or other measures domestically to give effect to the Convention’s provisions. At the multilateral level the Convention foresees a Conference of the Parties as a follow-up mechanism to monitor implementation, exchange good practices, coordinate cooperation and support mutual assistance. The Convention requires States to establish or adapt domestic oversight bodies, complaints and remedial mechanisms, and administrative arrangements necessary to ensure compliance with human-rights, democratic and rule-of-law obligations when AI systems are used by public authorities or private actors within the State’s jurisdiction.

Key Focus Areas

Fundamental principles: respect for human dignity and individual autonomy; equality and non-discrimination; privacy and personal data protection; transparency and oversight; accountability; reliability and safe innovation.
Risk- and lifecycle-based approach: iterative impact and risk assessment requirements across an AI system’s lifecycle and adoption of appropriate prevention and mitigation measures, including the possibility of restrictions, bans or moratoria for certain high-risk applications.
Transparency and procedural safeguards: documentation and record-keeping requirements; notice obligations for interactions involving AI; requirements to provide affected persons with information sufficient to challenge or seek review of AI-based outcomes.
Two-track approach to private-sector regulation: Parties may directly subject private actors to Convention obligations or achieve Convention objectives through alternative measures consistent with human-rights obligations.
Exemptions and scope limits: activities related to national defence or national security may be exempt where conducted in accordance with applicable international law and where democratic institutions and processes are respected; research and development activities are generally outside scope except where testing or similar activities risk interfering with rights, democracy or rule of law.
Cooperation and mutual assistance: obligations for Parties to cooperate, share information and provide mutual assistance to implement the Convention’s objectives effectively.

Implementation Framework

Implementation of the Convention is expected to occur through domestic legislative, administrative and other measures tailored by each Party to its legal system. Core implementation elements include:

Adoption of legislative, regulatory or administrative measures to reflect the Convention’s object and purpose across sectors and use-cases.
Establishment of iterative risk and impact assessment processes across the AI lifecycle, including mitigation plans and, where necessary, prohibition of certain uses.
Documentation and record-keeping obligations designed to enable contestability, oversight and effective remediation (audit trails, model cards, data provenance, testing and evaluation records).
Transparency obligations such as notice when interacting with automated systems and disclosure to affected persons sufficient to challenge or seek review of AI-driven decisions.
Creation or designation of oversight authorities, complaint-handling procedures and access-to-remedy processes to address harms and ensure administrative and judicial review where AI materially affects rights.
Mechanisms for cross-border cooperation, information sharing and mutual legal assistance among Parties to support enforcement and compliance.

Monitoring and Evaluation

The Convention foresees a structured follow-up and monitoring regime implemented through the Conference of the Parties and related intergovernmental mechanisms. Parties are expected to meet reporting obligations, participate in peer exchanges, and share good practices to monitor implementation progress. Monitoring and evaluation occur primarily at the domestic level (via national oversight bodies, reporting and administrative review) and at the intergovernmental level through the Convention’s follow-up machinery, which facilitates assessment of compliance, identification of implementation gaps and cooperative responses.

Penalties, Liability, and Appeals

The Convention requires Parties to ensure effective domestic remedies, oversight, complaint procedures and mechanisms for affected persons to seek redress where AI systems substantially affect rights. It does not itself establish a supranational enforcement court specific to the Convention; enforcement is primarily domestic through national legislative or administrative measures, courts and oversight authorities. The Convention’s intergovernmental monitoring, reporting and Conference of the Parties processes provide cooperative and peer-review pressures to promote compliance rather than a centralized supranational sanctioning mechanism. Parties retain responsibility to define penalties, administrative sanctions and liability rules within their domestic legal frameworks consistent with the Convention’s obligations.

Relationship to Other Instruments

The Framework Convention complements and aligns with other international and regional instruments that address AI, human rights and data protection. It explicitly centres human rights, democracy and the rule of law and is designed to work alongside non‑binding instruments such as the OECD AI Principles and other international standards. It also interacts with regional and sectoral instruments, including the EU AI Act and Council of Europe instruments on data protection (notably Convention 108). Parties should implement the Convention in a manner coherent with their obligations under other international treaties and regional legal frameworks.

International Alignment

The Convention is intended as a global instrument negotiated by Council of Europe member States together with several non-member States and the European Union. It was adopted by the Committee of Ministers on 2024-05-17 and opened for signature on 2024-09-05. Many States and the EU have signed the Convention; entry into force is conditioned on the deposit of a minimum number of ratifications. Uruguay’s signature on 2025-09-02 made it the first Latin American State to sign the treaty, signifying diplomatic alignment with the Convention’s human-rights-based approach and strengthening interoperability with European and other international approaches to AI governance.

Implementation Timeline

Date	Event
2024-05-17	Committee of Ministers of the Council of Europe adopted the Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law (CETS No. 225).
2024-09-05	Convention opened for signature (official opening for signature at Vilnius).
2025-02-25	Uruguayan executive resolution (entry dated 25/02/2025) designating plenipotentiary to sign the instrument (preparatory domestic act).
2025-04-23	Presidential resolution (Res. S/N/025, 23 April 2025) in Uruguay designating Ambassador Enrique Loedel as plenipotentiary to sign the Convention.
2025-09-02	Uruguay signed the Framework Convention in Strasbourg; Uruguay became the first Latin American State to sign the Convention (signature performed by Ambassador Enrique Emilio Loedel Soca).
—	Entry into force: the Convention enters 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 deposited instruments of ratification; signature alone does not equal ratification or entry into force.

Compliance Checklist

Requirement	Description
Legislative and administrative measures	Adopt domestic laws, regulations or administrative rules as appropriate to give effect to Convention obligations and ensure consistency with human-rights, democratic and rule-of-law standards.
Risk and impact assessments	Implement iterative lifecycle risk and impact assessment procedures for AI systems, including documented mitigation plans and escalation processes for high-risk applications.
Documentation and record-keeping	Maintain documentation enabling contestability and oversight (e.g., logs, model documentation, training data provenance, testing records) sufficient to audit and review AI-driven decisions.
Transparency obligations	Provide notices when interacting with AI; disclose to affected persons information sufficient to understand and challenge AI-based decisions where rights are affected.
Procedural guarantees and remedies	Establish complaint mechanisms, oversight bodies and access-to-remedy pathways that permit individuals to seek redress, including administrative and judicial review where appropriate.
Oversight and enforcement	Create or designate authorities with powers to monitor compliance, investigate breaches and enforce applicable domestic sanctions or corrective measures consistent with the Convention.
International cooperation	Participate in Conference of the Parties reporting, mutual assistance and information-sharing to support cross-border oversight and harmonised implementation.

Sources and References

Source	URL
Council of Europe — The Framework Convention on Artificial Intelligence (official treaty information and description)	https://www.coe.int/en/web/artificial-intelligence/the-framework-convention-on-artificial-intelligence
Council of Europe — Council news: Uruguay signs Council of Europe’s global AI treaty (press note)	https://www.coe.int/en/web/artificial-intelligence/-/uruguay-signs-council-of-europe-s-global-ai-treaty-1
Agencia de Gobierno Electrónico y Sociedad de la Información y del Conocimiento (Agesic) — Uruguayan government announcement	https://www.gub.uy/agencia-gobierno-electronico-sociedad-informacion-conocimiento/comunicacion/noticias/tratado-del-consejo-europa-sobre-ia-uruguay-es-primer-pais-latinoamerica
Presidency of Uruguay — Official resolution naming plenipotentiary to sign the Convention (Res. S/N/025, 23 April 2025)	https://www.gub.uy/presidencia/institucional/normativa/resolucion-sn025-se-designa-plenipotenciario-enrique-loedel-para-suscribir
Council of Europe — Treaty texts and explanatory materials (CETS No. 225; treaty text and treaty-series information)	https://book.coe.int/en/texts-of-council-of-europe-treaties/12225-council-of-europe-framework-convention-on-artificial-intelligence-and-human-rights-democracy-and-the-rule-of-law-cets-no-225.html

Uruguay has become the first Latin American country to sign the Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law, an international treaty designed to ensure AI systems respect fundamental human rights, democratic principles, and the rule of law. While Uruguay signed the Convention in September 2025, this diplomatic step does not yet make it legally binding for the country. The Convention itself needs to be ratified by at least five countries (including three Council of Europe members) to enter into force, and Uruguay would then need to ratify it domestically for its provisions to apply.

Once in effect for Uruguay, the treaty will require the government to implement domestic laws and measures that apply to both public authorities and private companies developing, deploying, or operating AI systems within its jurisdiction. The definition of an "Artificial Intelligence System" is broad, covering any machine-based system that infers from input to generate outputs like predictions or decisions that can influence physical or virtual environments.

Key obligations under the Convention include: - Conducting iterative risk and impact assessments throughout an AI system's entire lifecycle, from design and development to decommissioning. - Ensuring transparency through robust documentation, record-keeping, and providing clear notice to individuals when they interact with AI, giving them enough information to challenge or seek review of AI-based outcomes. - Establishing domestic oversight bodies, complaint mechanisms, and effective ways for individuals to seek remedies for any harms caused by AI systems. - Potentially restricting or even banning certain high-risk AI applications if necessary to protect rights.

The Convention does not set specific penalties. Instead, it requires Uruguay to establish its own domestic enforcement framework, including defining penalties, administrative sanctions, and liability rules. An international "Conference of the Parties" will monitor implementation and facilitate cooperation among signatory states. A practical pitfall for businesses is that while the specific national laws are still to be developed after ratification, this signing signals a clear future direction for AI regulation in Uruguay, emphasizing human-centric design and accountability.

Uruguay - AI and Human Rights (CETS No. 225)

Summary

Plain English