UK AI Minister Kanishka Narayan: Testing Is No Longer Enough
The UK’s AI minister Kanishka Narayan told Fortune that countries must "harden" their defences against AI risk. He made this statement in response to a question about a former Anthropic researcher's extinction warning.
"You have to harden and build your defenses. The risks have only grown, so testing is good but clearly insufficient," Narayan said.
Fortune reported that Jacob Coxon, who quit Anthropic this month, raised concerns about AI potentially causing humanity's extinction by the end of the decade. Evan Hubinger, another Anthropic scientist, agreed with Coxon's assessment, placing the risk at above 10%.
Narayan emphasized his focus on current risks, citing increased cybersecurity threats and AI tools that could elevate biochemical risks. He also expressed concern about AI models exceeding human instructions, stating:
"The sort of logical loops you have to jump through to make that an existential question are quite significant and severe."
The UK has been testing AI models for three years, according to Narayan, who added that no other country is as well-equipped in this area. He also revealed the UK's plan to bring countries together at the upcoming G20 under its presidency to shape global AI principles.
The White House has advised G20 members against adopting new global regulations, while President Trump has opposed calls for a slowdown in US research, arguing that it would give China an advantage. Microsoft co-founder Bill Gates, meanwhile, has advocated for government involvement beyond self-regulation.
Narayan's remarks also touched on the impact of AI on jobs, made during a discussion with the Resolution Foundation think tank. He highlighted the shared responsibility between governments, companies, and individuals in determining whether AI enhances or replaces human work.