Comment on this storyCommentAdd to your saved storiesSave
Policymakers in Brussels remained in late-night negotiations on Wednesday to reach an agreement on the world’s most ambitious law regulating artificial intelligence. Meanwhile, senators signaled that the U.S. Congress is taking a different approach to the new technology than the European Union, with lawmakers expressing concerns that the bloc’s approach could be heavy-handed and risk alienating AI developers.
The split-screen events on both sides of the Atlantic highlight the challenges of regulating artificial intelligence, a growing priority for governments around the world in the year since the release of AI-powered chatbot ChatGPT sparked global excitement.
The congress is far behind its counterparts in Brussels, where a framework to regulate AI was proposed for the first time in 2021. But after years of work, the future of the EU’s AI legislation remains uncertain in the face of a lobbying campaign and opposition from the EU’s largest nations – France, Germany and Italy.
Schumer launches “all hands on deck” push to regulate AI
After more than half a year of working on AI policy, Senate Majority Leader Charles E. Schumer (D-N.Y.) told reporters that the bipartisan group is “starting to really work on the legislation,” although he only Few details were given about it. Such a bill would include the following:
The comments came during Congress’ final two AI Forums of 2023, where lawmakers met with top technology executives, including former Google CEO Eric Schmidt, to better address issues such as the risks of an AI doomsday and national security understand.
Sen. Mike Rounds (R-S.D.), a member of the bipartisan working group Schumer assembled to craft AI policy, said senators were taking an “incentive-based” approach to keeping AI developers in the United States.
“If [European policymakers] “If we look at this as a regulatory activity, they will move AI development to the United States,” he told reporters after the two forums. “We don’t want to chase AI development after our opponents.”
Europe is making progress on AI regulation and challenging the power of the tech giants
Meanwhile, officials in the European Union sought a late breakthrough on the EU’s AI law, which would largely take a “risk-based” approach to limiting the use of AI applications based on how dangerous they are predicted by lawmakers.
European Parliament officials are scrambling to counter attempts by the 27-member bloc’s largest nations to weaken the historic bill. In recent weeks, discussions between the EU’s various bodies – the European Commission, which proposes laws, and the European Council and European Parliament, which pass them – have seen disagreements, threatening a law that has been years in the making. Officials entered the negotiations optimistic that a compromise could be reached, and talks were still underway as midnight approached midnight in Brussels.
If no agreement is reached in the marathon phase of negotiations, which is expected to last until Thursday morning Brussels time, negotiations would likely move to a final attempt in January, after which experts say it could be difficult before the European elections Parliament to pass a bill in June.
“If we go beyond January, I think we are lost,” said Brando Benifei, one of two lawmakers responsible for the law in the European Parliament. “It will take at least nine months before we could have the AI law.”
The EU’s largest nations have sought to scrap part of the bill that would impose mandatory regulations and transparency rules on foundation models, such as the underlying ChatGPT technology, which generates answers based on models created by scraping data from the trained throughout the Internet. Arguing that these rules could stifle innovation and put Europe even further behind the United States in the race to develop such models, these countries instead pushed for industry self-regulation.
People familiar with the talks, who spoke on condition of anonymity to describe delicate negotiations, said France appeared to be the strongest obstacle to an agreement, based in part on its desire to protect an emerging company that develops AI foundational models developed, Mistral based in Paris. as well as other French AI companies. An attempt to limit AI in policing, meanwhile, comes as France plans to use AI-powered smart cameras for policing and security at the 2024 Summer Olympics, and French cities have already entered legal gray areas by deploying or testing such technologies.
Asked about the French opposition, France’s Digital Minister Jean-Noël Barrot said that European governments were largely against restrictions on the use of AI for policing and national security and that burdensome regulations on foundation model developers could seriously hinder European innovation.
“There is unanimous consensus in the Council that the use of AI for national security purposes should not be included in the regulation,” he said.
He added: “The [AI] “Industry in Europe has expressed concern that placing too much burden on the shoulders of developers of basic models would be tantamount to not allowing these models to be developed in Europe,” he said.
Barrot insisted that even the compromise sought by the French would still result in the world’s strictest law regulating AI. He described the bill as the beginning and not the end of European regulations for the technology.
“I challenge anyone to present me with regulation as strict as the EU AI law around the world,” he said.
At the start of Wednesday’s negotiating session, Benifei said the push by France and other countries to allow industry to self-regulate would undermine one of the bill’s most important elements, arguing that a compromise must be found that imposes real restrictions.
“The most powerful models will be the foundation of all AI,” he said. “If we regulate their security and their transparency around how they work and the data used to train them, then we make it safer for all AI systems on the chain.”