EU Officials Reach Deal On ‘Historic’ AI Regulation

Topline Officials in the European Union reached a deal Friday on a sweeping bill aiming to regulate artificial intelligence in what is one of the world’s first major, comprehensive attempts to regulate the technology. In this photo illustration, the European Union (EU) logo and flag seen displayed on a smartphone Key Facts EU lawmakers reached agreement Friday on the key elements of the AI Act, which will be the first major regulation of AI technology that attempts to allow it to be used beneficially while still protecting against risks. The deal came after three days of negotiations in Brussels—including a 22-hour session stretching from Wednesday to Thursday, to the New York Times—and a copy of the agreed-upon act was not yet available.

Previous versions of the Act looked to limit how law enforcement, government and service companies, like water and energy providers, could use the technology, and would implement transparency requirements for major AI companies like OpenAI, Microsoft and Google. It would also require images made with AI to be labeled as such, something the United States is to implement, as well. The deal makes the EU the first governmental body to pass comprehensive AI regulation.

The deal does not have final approval as it still needs to be voted on by the 27 countries and technical details still need to be finalized, the New York Times reported, so it’s unclear when it will go to a vote. Crucial Quote “Historic!” European Commissioner Thierry Breton on X, formerly known as Twitter, on Friday. “The #AIAct is much more than a rulebook — it’s a launchpad for EU startups and researchers to lead the global AI race.

The best is yet to come!” What To Watch For What made it in the agreed upon draft. Reuters that going into discussions Wednesday, major countries like Germany and France were challenging proposals including the requirement that companies document their system’s training, show they’ve taken steps to mitigate risks and undergo audits, saying they would rather allow companies to self-regulate. What We Don’t Know How the AI Act will be enforced.

It will supposedly involve regulators from 27 nations and will require new hires, the New York Times reported, and is likely to face legal challenges from major tech companies before it goes into effect. Key Background The EU started work in 2018 on what would become the AI Act, setting itself up to be a lead regulator for the emerging technology. The first draft of it was available three years after work started, according to the New York Times, but it didn’t address “general-purpose” models like what is behind ChatGPT and Google’s Bard, because of how quickly the technology was developing.

The United States has been working to regulate AI, but Congress has yet to coalesce behind anything as comprehensive as the AI Act. The White House released an executive order in outlining “the most sweeping actions ever taken to protect Americans from the potential risks of AI” that required some companies to share results of safety testing and other information with the government, among other things. Further Reading.