Open Editor’s Digest for free
FT editor Roula Khalaf picks her favorite stories in this weekly newsletter.
EU lawmakers have agreed terms for a landmark law to regulate artificial intelligence, pushing ahead to enact the world’s most restrictive regime on the technology’s development.
EU Commissioner Thierry Breton confirmed a deal had been reached in a post on X, calling it a historic deal. “The EU has become the first continent to set clear rules for the use of AI,” he wrote. “The AI Law is more than a rulebook – it’s a launching pad for EU start-ups and researchers to lead the global AI race.”
The agreement follows years of debate among member states and European parliamentarians about the ways in which AI should be regulated to keep humanity’s interests at the heart of the law. This comes after marathon debates that began on Wednesday this week.
Details of the deal are still emerging after the announcement. Breton said legislators agreed on a two-tiered approach across the EU, with “transparency requirements for all general-purpose AI models (such as SatGBT)” and “stronger requirements for powerful models with systemic implications”.
Breton said the rules would avoid “overburdening” companies while implementing safeguards for the use of AI technology.
In the new rules, lawmakers agreed to stricter restrictions on the use of facial recognition technology with narrowly defined law enforcement exceptions.
The law also includes bans on using AI for “social scoring” — using metrics to establish how superior someone is — and AI systems that “manipulate human behavior to override their free will.”
Using AI to exploit vulnerable people due to age, disability or economic situation is also prohibited.
Companies that fail to comply with the rules will be fined up to €35mn, or 7 per cent of global revenue.
Some technical teams are not happy. Cecilia Bonfeldt-Dahl, director general of DigitalEurope, which represents the continent’s technology sector, said: “We have a deal, but at what cost? We fully supported a risk-based approach based on the applications of AI, not the technology, but a last-minute attempt to regulate the underlying models has reversed this.
“The new requirements — on top of other new laws like the Data Act — will take a lot of resources for companies to comply with, with resources being spent on lawyers instead of hiring AI engineers.”
MEPs argued their position for years before negotiations began with member states and the European Commission, the EU’s executive body. All three – national ministries, members of parliament and the commission – agreed to a final text on Friday night, allowing the legislation to become law.
European companies have expressed their concern that overly restrictive rules on the technology, which is growing and gaining traction after the popularity of OpenAI’s ChatGPT, will stifle innovation. In June, dozens of large European companies, including France’s Airbus and Germany’s Siemens, said the rules were too tough to foster innovation and help local industries.
Last month, the UK hosted a summit on AI security, leading to broad commitments from 28 countries to tackle existential risks arising from advanced AI. The event attracted leading tech figures such as OpenAI’s Sam Altman, who had previously criticized the EU’s plans to regulate the technology.
Ursula von der Leyen, president of the European Commission, the EU’s executive body, praised lawmakers for their political agreement on AI rules.
He said: This is a historic moment. The AI Act ushers European values into a new era.
He added: “We will support businesses and developers to anticipate the new rules until the legislation is fully implemented. Around 100 companies have already expressed their willingness to join our AI Pact, thereby pledging on a voluntary basis to implement key obligations of the Act ahead of the statutory deadline.