UK AI minister urges stronger defenses against advanced AI risks
U.K. AI minister Kanishka Narayan said warnings from people inside AI labs about severe risks from advanced systems must be taken seriously and that countries need stronger safeguards. He argued that testing alone is insufficient and said the U.K. wants to use its G20 presidency to help shape international principles. The remarks come as U.S. officials resist strict global AI regulation while some lawmakers push for new oversight of frontier labs.
Narayan, a Cabinet minister, spoke at the U.K. Labour Party’s annual gathering. He cited warnings from former Anthropic staffer Jacob Coxon, whose X post drew 175 million views, and from Anthropic’s Evan Hubinger, who put the danger above 10%. Bill Gates separately said misuse could contribute to events killing a billion people.
Narayan said Britain aims to use next year’s G20 presidency to bring nations together and help set shared principles. He listed cybersecurity, biochemical proliferation and models acting beyond instructions as concerns, while noting AI’s potential to create jobs and raise productivity. Washington has reportedly urged G20 members to avoid fresh international rules.
The debate may shape how quickly frontier AI is deployed and how much oversight labs face. Workers could see productivity gains or job displacement depending on adoption choices. Citizens may benefit from stronger cybersecurity and biosecurity safeguards, though stricter rules could also slow innovation or widen gaps between countries. International coordination may influence trust and competitiveness, but outcomes remain uncertain.