Parliament backs EU signature of Framework Convention on Artificial Intelligence

First international legally binding treaty on artificial intelligence (AI) governance

Reinforces existing EU legislation, such as AI Act

Includes rules on transparency, documentation, risk management, oversight, and a risk-based approach

On Wednesday, Parliament approved the Council of Europe’s Framework Convention on AI, tackling risks AI poses to fundamental democratic values at international level.

With 455 votes in favour, 101 against, and 74 abstentions, Parliament gave its consent to the EU’s signature of the Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law. The convention embeds existing EU policies on artificial intelligence in an internationally recognised framework to support the safe, rights-respecting deployment of AI across the EU and worldwide.

The convention seeks to ensure that AI systems adhere to strict ethical standards throughout their lifecycles. It promotes transparency, auditability and effective oversight, and strengthens safeguards to protect fundamental rights, democracy and the rule of law.

Implementation

The framework convention is the first legally binding international treaty dedicated specifically to AI governance. It applies to AI-related activities carried out by public authorities, as well as by private actors acting on their behalf. Private sector actors remain responsible for addressing AI-related risks in line with the convention’s objectives. They may either apply the convention’s obligations directly or achieve equivalent protection by other means.

In practice, the convention establishes a global baseline for AI governance. Within the EU, the AI Act and other EU legislation already set a higher, more detailed level of protection and harmonisation within the internal market.

The convention is consistent with EU law, including rules governing the placing of AI systems on the market. Its principles align with existing EU legislation, such as:

the AI Act, which sets binding requirements for data governance, cybersecurity, transparency and monitoring, and bans certain unacceptable AI practices;

EU data protection rules (GDPR), which safeguard personal data through risk-based protections;

EU non-discrimination law, which prohibits discrimination and addresses data quality and bias mitigation in high-risk AI systems;

sector-specific legislation on product safety, liability, and political advertising.

In their position, co-rapporteurs José Cepeda (S&D, ES) and Paulo Cunha (EPP, PT) stress that the conclusion of the framework showcases the EU’s commitment to safe development and deployment of AI. By leading this technological revolution with democratic and human-centric values at its core, the EU is demonstrating that AI must serve people, strengthen open societies, and uphold a European model founded on human dignity, transparency, and accountability. This approach also supports stable and safe economic growth, a path that the Parliament fully endorses.

Next steps

Now that Parliament has endorsed the Commission’ signature to become a party to the convention, Council can conclude the agreement.

Background

Negotiations on the convention began in September 2022 under the auspices of the Committee on AI (CAI) established by the Council of Europe in Strasbourg. The European Commission, on behalf of EU, negotiated alongside Council of Europe member states, and a range of international partners, including the United States, Canada, Mexico, and Japan. Civil society, academia, industry and international organisations also contributed.

The convention is open to EU member states and partners worldwide. In addition to the European Union, current signatories include the United Kingdom, Ukraine, Canada, Israel and the United States.