By Jillian Deutsch
The European Union has reached a hard-fought agreement on what is poised to become the most comprehensive regulation of artificial intelligence in the Western world.
Thierry Breton, head of the bloc's internal market, said the deal strikes a balance between promoting innovation and protecting the rights of people and businesses.
“We have spent a lot of time finding the right balance between maximizing the potential of AI to support law enforcement while protecting the fundamental rights of our citizens,” he said in a statement early Saturday. “We don't want mass surveillance in Europe.”
After more than 37 hours of negotiations this week, representatives from the European Commission, the European Parliament and 27 member states agreed to a series of controls for generative artificial intelligence tools such as OpenAI Inc.'s ChatGPT. and Google's Bard – the kind capable of producing content on command.
Deal!#AIAct pic.twitter.com/UwNoqmEHt5
— Thierry Breton (@ThierryBreton) December 8, 2023
Negotiators agreed to allow live facial scanning, but with safeguards and exemptions, Breton said. The deal would ban biometric scanning that categorizes people based on sensitive characteristics such as political or religious beliefs, sexual orientation or race. Officials said this was one of the most difficult and sensitive issues in the talks.
The proposed legislation would impose financial penalties on companies that break the rules, with fines of up to €35 million, or 7% of global turnover, depending on the violation and the size of the company.
The bill still needs to be formally approved by the EU member states and parliament. But the deal marks a crucial step toward a groundbreaking AI policy that — in the absence of any meaningful action from the U.S. Congress — will set the tone for regulation of the rapidly developing technology. The EU aims to establish the first strong AI guardrails outside Asia.
Policymakers have worked for months to finalize the language in the AI law and get it passed before June's European elections usher in a new commission and parliament that could force more changes and delay the effort.
The decision was made during a session on Friday after a nearly 24-hour marathon that stretched from Wednesday to Thursday. During the first meeting, some negotiators in the room dozed off while others debated the most sensitive topic: restricting live facial scanning technology in public, before finally agreeing to break it.
The difficult discussions underscore how contentious the debate over regulating AI has become, dividing global leaders and tech executives alike as generative tools become increasingly popular. The EU – like other governments including the US and UK – has struggled to balance the need to protect its own AI startups, such as France's Mistral AI and Germany's Aleph Alpha, from potential social risks.
That proved to be a major sticking point in the negotiations, with some countries, including France and Germany, opposing rules they say would unnecessarily hinder local businesses.
Some details will still be worked out by officials in the coming weeks, but negotiators have largely agreed to set rules around generative AI that include basic transparency requirements for any developer of a major language model. Those who pose a systemic risk will be required to sign a voluntary code of conduct to work with the commission to mitigate risks. The plan is similar to the EU's content moderation rules, the Digital Services Act.
The most difficult point came down to the question of how much to restrict live biometric identifiers. Parliament voted for a full ban last spring, but EU countries pushed for exemptions for national security and law enforcement. Ultimately, the two sides agreed to limit the technology's use in public spaces with more guardrails.
Both Parliament and the Council, to which the EU's 27 member states belong, will have to approve the agreement. France and Germany were previously highly critical that the law could overregulate general-purpose AI systems and kill domestic OpenAI competitors such as Mistral and Aleph Alpha.
Carme Artigas, Spain's foreign minister, said the deal falls within the scope of what EU countries wanted, and “we hope they will all confirm this.” She also added that Mistral is unlikely to have to comply with general purpose AI controls for now as the company is in the research and development phase.
“We will carefully analyze the compromise found today and ensure in the coming weeks that the text preserves Europe's ability to develop its own artificial intelligence technologies and its strategic autonomy,” said French Digital Minister Jean-Noel Barrot , in a statement.