The AI Act, a comprehensive rulebook governing artificial intelligence, has achieved unanimous approval from all 27 EU member states. Following extensive negotiations, ambassadors representing these countries within the bloc have endorsed the text.
(Photo : KENZO TRIBOUILLARD/AFP via Getty Images)
EU commissioner for internal market Thierry Breton speaks during a press conference with European Commission vice-president in charge for alues and Transparency Vera Jourova, on the Media Freedom Act at the EU headquarters in Brussel, on September 15, 2022.
Achieving Unanimous Approval
This marks a significant step toward regulating the deployment and use of artificial intelligence across the European Union, as reported by The Verge.
In December, European Union policymakers successfully navigated through key points of contention regarding the AI Act, a landmark legislative initiative aimed at regulating Artificial Intelligence's impact on potential harm.
The intricate nature of the law necessitated over a month of technical refinement. On January 24th, at a technical meeting, the Belgian presidency of the Council of EU Ministers revealed the final version of the text.
Nevertheless, lingering reservations prevailed among many member states, stemming from inadequate time for a thorough analysis of the comprehensive text.
Paris, Berlin, and Rome initially had reservations about the provisional agreement. As member states, particularly France, tried to influence the law's preamble during technical discussions, the Belgians (now holding the presidency) limited concessions.
Despite France's attempts to delay, Berlin's early support and Italy's decision not to oppose played crucial roles. Internal disagreements in the French government were resolved, leading to France supporting the text with strict conditions.
Also Read: AI, Now Officially Considered as Potential Financial System Risk by US Regulators
Euractiv reported that Paris is advocating for the thoughtful implementation of the AI Act to foster competitive AI models, striking a balance between transparency and safeguarding trade secrets.
They emphasize the need to avoid imposing excessive obligations on companies and call for a reevaluation of the criteria for designating AI models with systemic risks.
Slovakia has expressed its concerns, seeking clarifications on crucial terms, advocating for international alignment, and exploring the possibility of EU countries regulating AI in non-professional activities.
Austria has presented a protocol statement, raising apprehensions about data protection and consumer law, particularly regarding law enforcement exceptions and regulations related to invasive technologies like remote biometric identification.
EU member states still wield influence over the AI law's implementation, with the Commission expected to issue approximately 20 acts of secondary legislation. The AI Office, responsible for overseeing AI models, will be notably staffed with seconded national experts.
What's Next?
The AI rulebook is set for adoption by the European Parliament's Internal Market and Civil Liberties Committees on February 13, with a plenary vote scheduled for April 10-11. Following these steps, Reuters reported that this will undergo formal adoption at the ministerial level.
The AI Act will become effective 20 days after being published in the official journal. Prohibitions on certain practices will take effect six months later, while obligations related to AI models will commence after one year.
The remaining rules will be enacted after two years, with the exception of the classification of AI systems requiring third-party conformity assessment under other EU regulations as high-risk, which has been delayed by an additional year.
Related Article: EU Warned by Experts, AI Models Could be Over-Regulated "Out of Existence"
