LONDON — Bailey Says AI Testing, Not Regulation, Should Come First

Bank of England Governor Andrew Bailey says regulating artificial intelligence "is not the right place to start," arguing instead that rigorous testing to find vulnerabilities and build safeguards must come first, according to BBC News.

Bailey made the case in his first-ever Substack post, published Tuesday, where he described the risks around AI as "real and increasingly significant" while stopping short of calling for a halt to development. "On the contrary, the benefits are immense," he wrote, according to the BBC, adding that a system for intervention and clear operating boundaries is still needed.

What Does Bailey Want Instead of Regulation?

Bailey pointed to testing as the immediate priority, not rulemaking. He said the UK's AI Security Institute is already doing "important work" that should feed into a broader set of standards, but cautioned that "the pace of progress must accelerate." He added a caveat: "We should proceed with a degree of humility," according to the BBC report.

Why Frame Model Failures as Useful?

Bailey argued that unexpected model behavior during testing should not be read as a setback. "That is not evidence that testing has failed, rather it is evidence of why testing is necessary," he said, per the BBC. He singled out so-called frontier AI — advanced, self-learning systems — warning that without effective intervention tools, such models risk becoming "a closed loop in which the model progressively governs itself."

"That is not evidence that testing has failed, rather it is evidence of why testing is necessary." — Andrew Bailey, Governor, Bank of England

Does Bailey Rule Out Future Regulation?

No. Bailey said testing should not be treated as a substitute for regulation and that a more formal framework could emerge over time. His position is that testing comes first, not that rules never come — a distinction he drew explicitly in the piece, per the BBC.

How Does This Compare With the White House Approach?

Bailey's testing-first stance lands alongside a separate move in Washington. President Donald Trump this week hosted executives from OpenAI, Anthropic, Nvidia, SpaceX, Meta and Google, after which he said the companies had signed a "morally binding" document intended as a "form of protection" against AI's risks, the BBC reported. Under that arrangement, each company remains responsible for the safety of its own technology — a self-policing model that differs from Bailey's call for a UK testing regime tied to a national institute. Trump has separately argued against slowing AI development, saying the US is racing China and adding, "whoever wins AI, wins," according to the BBC.

What Are AI Companies Themselves Saying?

The governor's remarks follow weeks of intensifying debate inside the industry itself. Leaders at Anthropic and OpenAI have called for slower development and internationally coordinated risk assessment, and OpenAI recently shelved release of its newest model over safety concerns, the BBC noted. That tension — between firms urging caution and a US administration pushing speed — is playing out alongside investor debate over AI's economic payoff, a split detailed in HTT News' coverage of JPMorgan's AI outlook.

Bailey did not set a timeline for when testing-based standards might mature into formal rules.