UK banks still lack access to Mythos AI model, Bank of England governor says

Bank of England Governor Andrew Bailey said UK banks remain unable to access Anthropic’s Mythos AI model for cybersecurity testing six weeks after initial concerns were raised. Speaking to Bloomberg TV during a central banking conference in Reykjavik, Bailey attributed the delay to a political process involving the U.S. administration, though he could not explain why approval varies between companies. He stressed that access to the model is critical due to the significant cyber risks it poses. Anthropic has faced tensions with the U.S. government over military use restrictions for its AI tools, which Bailey previously described as potentially ‘cracking the whole cyber risk world open.’ While some cybersecurity experts have downplayed fears of unrestricted hacking risks, Bailey warned that financial institutions cannot operate in isolation. He noted that the interconnected nature of global banking means a single national approach to cybersecurity is insufficient. Bailey, who also leads the Financial Stability Board, called for a coordinated international strategy to address cyber threats. He stated that spillover effects from cyber risks are too vast to be contained by individual countries, urging collaboration beyond domestic borders. The delay in accessing Mythos highlights broader challenges in aligning AI governance across nations. Meanwhile, U.S. President Donald Trump postponed signing an executive order aimed at creating a voluntary framework for AI developers to engage with the government before releasing advanced models. The move further complicates efforts to standardize AI oversight, leaving regulatory gaps that Bailey described as dangerous for global financial stability.