top of page

EU Ambassadors Unanimously Approve Landmark AI Rulebook

EU ambassadors have unanimously approved the first comprehensive AI rulebook, setting a new global standard in AI regulation.


EU ambassadors have unanimously approved the first comprehensive AI rulebook.
Credit: Christian Lue, Unsplash

BRUSSELS, BE – In a groundbreaking development, the ambassadors of the 27 European Union countries have unanimously approved the world's first comprehensive rulebook for Artificial Intelligence. This monumental approval, reached on 2 February, rubber-stamps the political agreement forged in December and marks a significant milestone in global AI governance.


December's political agreement on the AI Act, a flagship bill by EU policymakers, aimed to regulate AI based on its capacity for harm. The complexity of the law necessitated over a month of technical refinement, culminating in the Belgian presidency of the Council of EU Ministers presenting the final text on 24 January. Initially, member states harbored reservations, citing inadequate time to analyze the text in its entirety.


These reservations were lifted with the adoption of the AI Act from the Committee of Permanent Representatives. Despite initial uncertainties, especially from major European countries like France, Germany, and Italy, the agreement was reached.


Pushback And Compromise


France, a primary opponent of the initial agreement, along with Germany and Italy, had sought a lighter regulatory regime for powerful AI models, such as OpenAI's GPT-4. They argued for limiting regulations to codes of conduct, to foster the growth of European AI startups like Mistral AI and Aleph Alpha, in competition with American companies.


The European Parliament, however, advocated for stringent regulations, particularly for high-capacity AI models, to ensure a fair regulatory environment. The eventual compromise adopted a tiered approach: horizontal transparency rules for all AI models and additional obligations for models posing systemic risks.


As the Belgian presidency succeeded the Spanish presidency in January, they resisted significant textual changes to preserve the agreement with the EU Parliament. France's attempts to delay the ambassadors' vote and push for concessions were ultimately unsuccessful, particularly after Germany's support of the text.


Italy, lacking a major AI startup to defend, also chose not to oppose the AI Act. This decision was influenced by its role in the rotating presidency of the G7, where AI is a key topic. Within the French government, the Ministry of Economy's opposition was counterbalanced by the Ministry of Culture and the Ministry of Interior, leading to France's conditional support of the text. In addition to France, Slovakia and Austria raised concerns, with Slovakia seeking clarifications on key terms and Austria voicing data protection and consumer law issues.


AI Regulation Implementation


The AI Act's implementation leaves room for influence by EU countries, as the European Commission will issue approximately 20 secondary legislation acts. The AI Office, responsible for overseeing AI models, will be significantly staffed with seconded national experts.


The European Parliament's Internal Market and Civil Liberties Committees are set to adopt the AI rulebook on 13 February, with a formal adoption expected after a plenary vote in April.

The AI Act will be enforceable 20 days post-publication in the official journal.


The prohibition of certain practices will be effective after six months, and obligations for AI models will commence after one year. Most regulations will be operational after two years, except for the classification of certain AI systems under EU rules as high-risk, which has been extended by an additional year.


This unanimous decision by the EU ambassadors not only sets a new precedent in AI legislation but also underscores the EU's commitment to balancing innovation with ethical and societal considerations in the burgeoning field of artificial intelligence.

bottom of page