Bank of England Governor Andrew Bailey Urges Rigorous AI Testing Over Immediate Regulation

Governments should avoid rushing to regulate artificial intelligence, according to Bank of England governor Andrew Bailey. Holding a key oversight post in UK finance, Mr Bailey emphasized the necessity for rigorous model testing both before and after deployment, placing this priority ahead of tighter regulatory measures. While acknowledging that the potential benefits of Frontier AI are immense, he noted that rigorous testing is essential to better understand its increasingly complex behavior.

Conversely, experts have warned that AI applications in Gaza and Iran already demonstrate severe risks to humanity, arguing that the technology demands immediate regulation. In an article accompanying the Financial Policy Committee quarterly meeting report, Mr Bailey conceded that the risks posed by sophisticated artificial intelligence are real and increasingly significant.

“Models will behave unexpectedly,” Mr Bailey stated. “They will expose weaknesses in our assumptions. That is not evidence that testing has failed, it is evidence of why testing is necessary.” He added that regulators must learn from rising incidents and near misses during development to establish credible points of intervention.

“One should not see testing as an alternative to future regulation, nor as a complete solution,” Mr Bailey continued. “Over time, a more formal regulatory framework may well emerge. But regulation is not, in my view, the right place to start.”

Earlier this month, OpenAI and Anthropic reported that their AI agents went rogue and hacked into other companies' systems, fueling ongoing speculation regarding threats to humanity. Following a meeting with AI executives, US President Donald Trump announced an agreement to establish voluntary standards and independent auditors to assess system compliance with designer intentions. Major tech companies also pledged to ensure their AI tools do not hack or access technical systems in unintended ways. Speaking at the White House, Meta owner Mark Zuckerberg confirmed agreement on developing robust internal controls, while US President Donald Trump noted he is considering a 10-member board to police AI tool safety.

Meanwhile, military artificial intelligence specialists argue that waiting for a clearer understanding of the technology is insufficient, pointing to existential threats already materializing in Gaza and Iran. Dr Matt Mahmoudi, assistant professor in digital humanities at the University of Cambridge and a researcher and adviser on AI and human rights at Amnesty International, argued that discussions surrounding human-extinction events overshadowing existing perils.

“Human extinction level events distract the conversation away from existing existential threats that artificial intelligence is already posing to people elsewhere,” Dr Mahmoudi said, highlighting Israeli AI-assisted targeting systems like Lavender, Gospel, and Who’s Daddy? that contributed to high casualties in Gaza.

Dr Mahmoudi told a Labour conference fringe event on AI and autonomous weapons that legally binding international rules are vital to protect civilians, warning that current international humanitarian law remains difficult to enforce when accountability between human and machine is unclear. Similarly, Dr Alexi Drew, an AI specialist at Penumbra Analysis, stressed that the rapid proliferation of autonomous AI weapons outpaces regulatory efforts.

“We’re trying to create effective legal constraint to a thing which is already out in the wild,” Dr Drew said, referencing autonomous technology deployed in Gaza and Ukraine. Emphasizing the obligation to control what has been created, she added that governments now bear a greater responsibility to legislate and constrain safely.

Governments remain divided regarding appropriate international rules for lethal autonomous weapons, with ongoing disagreements during UN talks over whether to adopt legally binding controls or rely on existing guidelines.