Europe edges closer to landmark AI regulations after France agrees to ratify AI Act
France was seeking additional concessions to safeguard its own AI startups
France has finally agreed to ratify the EU's AI Act after months of resistance, marking a significant breakthrough in the long-standing dispute over the regulation of AI technologies in Europe.
On Friday, the draft regulation received unanimous endorsement from all 27 member countries during a meeting of EU ambassadors in Brussels.
Originally proposed by Brussels in 2021, the EU's AI Act stands out for its comprehensive approach, encompassing binding regulations covering aspects from risk assessment to copyright protection. Notably, the law mandates that AI systems be evaluated based on their potential risks, with higher-risk applications subject to stricter obligations.
While the Act aims to address concerns surrounding the ethical and societal implications of AI deployment, it has also raised apprehension about potential repercussions on innovation within the EU.
Observers fear that the strict regulations could hamper the development of AI technologies, potentially placing EU firms at a disadvantage compared to their counterparts in the US and China.
France falls into line
Prior to the agreement on Friday, key EU economies, including Germany and France, expressed reservations about certain provisions of the AI Act.
France sought additional concessions to safeguard its own AI startups and protect the transparency of "base models" used in AI development. In particular, Germany and France voiced fears that the regulatory burden could impede the growth of the EU's emerging AI champions, such as Mistral in France and Aleph Alpha in Germany.
Sources told Reuters that behind the scenes, French AI startup Mistral, founded by former researchers from Meta and Google AI, along with Germany's Aleph Alpha, actively lobbied their respective governments on the matter. Meanwhile, Austria raised concerns regarding data protection measures.
The approval last week was finally assured after France, the final holdout, withdrew its opposition to the AI Act.
To address their concerns, the EU offered assurances to member states regarding the implementation and enforcement of the AI Act.
The creation of the EU's AI Office, tasked with overseeing compliance with the legislation, served as a pivotal concession.
France expressed satisfaction with the final text, citing a balanced approach to protecting copyright while fostering technological advancement.
Thierry Breton, the EU's leading figure in technology regulation, lauded the historical significance of the new rules.
Industry reaction
In contrast to the politicians' optimism, industry representatives remain cautious.
The Computer & Communications Industry Association (CCIA), representing tech giants like Google, Apple, Amazon and Meta, cautioned about potential hurdles ahead.
CCIA Europe's senior policy manager, Boniface de Champris, highlighted the persisting ambiguity in the new AI rules, expressing concerns that they could impede the development and deployment of innovative AI applications in Europe.
The next crucial steps for the AI Act involve a vote by a key committee of EU lawmakers scheduled for 13th February, followed by a vote in the European Parliament either in March or April before formal approval in May.
While certain provisions will come into effect within six months, the full implementation of the law is slated for two years later, allowing stakeholders time to adapt to the new regulatory landscape.
Commenting on the latest development on EU AI Act, Victor Botev, CTO of Oslo-based scientific AI start up Iris.ai, said:
"The tiered approach strikes a necessary compromise between blanket laws and tailored exceptions. By scaling requirements proportional to societal risks, the Act allows for distinctions between applications without hampering innovation.
"However, continued efforts are needed to ensure all voices are heard in this consultative process -especially startups and smaller AI developers pioneering new technologies."
Bruna de Castro e Silva, AI governance specialist at governance toolmaker Saidot, said that the EU green lighting the final text of the EU AI Act shows that "Europe is ready to set a responsible pace of innovation for AI."
He added that regulators need to stand firm against lobbyists.
"We must stay vigilant in this process to prevent dilution of the EU AI Act's original intent. It's vital to make sure that corporate stakeholders are able to understand the reason for the EU's policies and what they can do to follow them.
"What we need is transparent guidelines, interweaving governance with the development, deployment and scaling of AI systems, taking the kind of multidisciplinary and cross-sectorial approach it takes to make real change."