TL;DR:

• Anthropic will expand Claude Mythos access to UK financial institutions within the next week, having previously restricted the model to a handful of US technology partners.
• Bank of England governor Andrew Bailey, ECB president Christine Lagarde and Canada’s finance minister all flagged Mythos as a serious regulatory challenge at last week’s IMF spring meetings in Washington.
• UK regulators are expected to raise Mythos risks with bank chief executives and Whitehall officials in the coming weeks, tightening the feedback loop between capability release and supervisory response.

Pip White, Anthropic’s head of UK, Ireland and Northern Europe, told Bloomberg TV that engagement with UK bank chief executives has been “significant” since the model’s controlled release, and that the company expects to extend access to British financial institutions imminently. Until now, Mythos has been limited to a small group of US firms including Amazon, Apple and Microsoft.

Governance scramble at the IMF meetings

Finance ministers and central bankers used the IMF and World Bank spring meetings to compare notes on Mythos, which Anthropic itself has described as capable of finding and exploiting software vulnerabilities beyond “all but the most skilled humans”. Canadian finance minister François-Philippe Champagne called the model an “unknown unknown” and cast it as a higher-order threat than conventional geopolitical risk. ECB president Christine Lagarde acknowledged that no governance framework currently exists to contain a tool of this reach.

Andrew Bailey, who chairs the Financial Stability Board, framed the core policy tension directly: regulate too early and the UK risks distorting the technology’s development; regulate too late and the situation escapes control. His public remarks hint at the dilemma behind the Bank of England’s own shift toward live stress-testing of AI systemic risk, rather than further warnings.

UK context and what comes next

The UK access announcement lands alongside the £500 million Sovereign AI Fund, the FCA’s AI Live Testing initiative and the BoE’s AI-herding simulations — each a different lever on the same problem of how to absorb frontier capability safely. US Treasury secretary Scott Bessent summoned American bank bosses to Washington last week for an equivalent conversation; British regulators are expected to follow with UK systemically important firms in the coming weeks.

Looking forward

For UK bank technology teams, the immediate question is whether Mythos arrives as a security asset (patching legacy vulnerabilities faster than attackers) or an inbound threat that outruns their patching cadence. The Financial Times reports AI-enabled cyber attacks rose 89 per cent in 2025 with breakout time falling to 29 minutes — a baseline against which any Mythos-era improvement or deterioration will be measured in the next quarter.