Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
Architects of the EU’s landmark artificial intelligence act have urged Brussels to halt “dangerous” moves to water down the rules, which would spare big US tech groups such as OpenAI and Google from key elements of the law.
The European Commission, the bloc’s executive arm, is holding discussions that could result in more parts of the law — considered the world’s strictest regime regulating the development of AI — becoming voluntary rather than compulsory.
This includes provisions designed to force AI companies ensure cutting-edge models do not produce violent and false content or be used in election interference. The move comes following intense lobbying by Donald Trump and Big Tech giants who are challenging European efforts to regulate the development of AI.
Adhering to such demands is “dangerous, undemocratic and creates legal uncertainty,” according to a letter to the commission’s digital chief Henna Virkkunen from several prominent members of the European parliament involved in AI regulation.
“If providers of the most impactful general-purpose AI models were to adopt more extreme political positions, implement policies that undermine model reliability, facilitate foreign interference or election manipulation, contribute to discrimination, restrict the freedom of information or disseminate illegal content, the consequences could deeply disrupt Europe’s economy and democracy,” the MEPs wrote.
The signatories of the letter include most MEPs who negotiated the AI Act and Carme Artigas, the former Spanish minister for digitalisation and AI who led negotiations on behalf of EU member states when the final version of the AI Act was agreed in December 2023.
The EU’s AI Act classifies the technology into three categories based on the risk it poses to human health and security, for instance if it has medical applications or is being used in public transport. The higher the risk category, the bigger the reporting requirements. Powerful AI models also face extra obligations, such as having to be more transparent about how models are trained.
The current debate surrounds the drafting of a “code of practice” which will provide guidance to AI companies on how to implement the act that applies to powerful AI models such as Google’s Gemini, Meta’s Llama and OpenAI’s GPT-4.
The code of practice is being drafted by a group of experts, including Turing-prize winner Yoshua Bengio. A final version is set to be adopted by the commission in May. The experts have tried to find a tricky balance between ensuring the law has force while also ensuring AI companies sign up to it, according to people briefed on the drafting.
Brussels has faced intense US lobbying over the AI Act. Meta’s head of global affairs Joel Kaplan warned a Brussels audience in February that the code of practice would impose “unworkable and technically unfeasible requirements”.
Meta has also said it cannot ship multimodal large language models and its latest AI assistant in the EU due to the bloc’s privacy rules. Other US tech companies such as Google, as well as European groups such as Spotify and Ericsson have also been critical about the EU’s regulation on AI.
Meanwhile, US vice-president JD Vance used a speech at France’s AI Summit in Paris last month to hit out against “excessive regulation of AI” and warned that “AI must remain free from ideological bias”.
The new commission, which started its current mandate in December, has made clear it wants to focus more on attracting AI investment and has announced it was withdrawing a planned AI liability directive as part of a broader push for deregulation.
Speaking at a Financial Times event on Tuesday, Virkkunen said it was important that the code of practice “is really helping industries and stakeholders and SMEs, really giving guidelines, not setting more obstacles or reporting obligations”.
At the same time, she stressed that the commission remains committed to its main principles. “We want to make sure that Europe has a fair, safe and democratic environment also when it comes to the digital world.”