U.S. tech boosters love to ridicule our Continental cousins' pearl clutching on AI. But they may have some ideas worth studying.
As Washington and Silicon Valley scramble to rein in artificial intelligence, allies across the Atlantic are already having at it. Not a shock, to be sure. The European Union has typically raced ahead regulation-wise with the 2024 EU AI Act.
"The EU is not waiting for the AI car to crash," says Georgina Kon, a London-based privacy and technology partner at law firm Linklaters.
Meanwhile, Prime Minister Andy Burnham just pitched the United Kingdom as an "honest broker" on global AI oversight during a speech at the United Nations General Assembly.
But the EU doesn't have an off-the-shelf solution for California-bred AI agents. "The EU AI Act doesn't talk about pacing or pause in development," says William Echikson, senior fellow on tech policy at the Center for European Policy Analysis. "It talks more about transparency."
The U.K. has so far favored a consultative approach centered on its AI Security Institute, a government think tank committed to "building the world's leading understanding of advanced AI risks and solutions."
"AISI probably has the best testing team in the world," says Janet Haven, executive director of the Data & Society think tank in New York. "But they are not monitoring the labs and safety culture."
The European systems have influence, nonetheless, and not only because the EU and U.K. economies together are 80% of the U.S.'s size. The EU Act focuses on "high risk" AI applications. Some are banned outright, like "social scoring" or police cameras storing "biometric information." Others, like those that manage critical infrastructure, education, or job recruitment, require specific documentation and "data governance."
This approach has informed active or pending legislation in U.S. states including California and Colorado, Haven says.
The AISI's U.K. parent ministry cemented a partnership last year with Silicon Valley powerhouse Google DeepMind to "explore high-impact solutions for responsible AI adoption across the public sector."
Not everyone is a fan. The AI Act, with its still-evolving thicket of paperwork, "certainly doesn't encourage developers to use AI to create companies," Echikson says.
The political drift since the notorious Hugging Face hack, by rogue AI agents, is toward a heavier regulatory hand, however.
A cross-party group of European parliamentarians just called for strengthening product liability strictures on AI products-a potential dose of kryptonite for the AI bros and the companies they run, Data & Society's Haven says. "Fines can't deter powerful companies, but they are terrified of product liability," she adds.
The U.K. Parliament's Joint Committee on Human Rights called, in a Sept. 14 report, for "a single independent AI oversight body" that would enforce "more demanding obligations for higher risk AI systems and models." Polling shows that nine in 10 citizens support independent regulation, says Gaia Marcus, director of the Ada Lovelace Institute in London.
President Donald Trump and China's Xi Jinping, leaders of the AI superpowers, showed little interest in slowing down at their summit this week. But The EU and U.K. can still press the issue, with global public opinion, not to mention industry (a bit), on their side, Marcus argues. "We do have leverage in granting access to our markets," she says.
A little pearl clutching may not be such a bad thing after all.
Write to editors@barrons.com