OpenAI Offers UK Banks a Cyber AI Alternative After Anthropic Locks Them Out

OpenAI has offered nine major UK banks access to its cybersecurity AI tool, GPT-5.5 Cyber, BBC Business reported Monday. The move follows Anthropic blocking those same institutions from previewing its competing product, Claude Mythos, leaving a significant gap in the UK financial sector’s access to cutting-edge cyber AI.

Anthropic Shuts UK Banks Out of Mythos Previews

Both tools are engineered to uncover hidden vulnerabilities buried inside complex digital systems. Independent testing by the AI Security Institute found the two products delivered broadly comparable performance on assigned tasks. Anthropic made waves in April when it launched Mythos, claiming the model had unearthed a security flaw in a legacy system that had gone undetected for nearly three decades. Despite that headline-grabbing claim, UK banks have remained locked out of any preview access.

Bank of England Governor Andrew Bailey publicly flagged the access problem last week, warning that domestic lenders were unable to use Mythos to audit their own systems and applications. Several finance ministers, central bankers, and senior financiers have separately voiced alarm that Mythos could destabilise the broader financial system if widely misused.

A Long History of Legacy Risk in UK Finance

The concern over old code is not new. Surrey University cybersecurity professor Alan Woodward told the BBC that portions of the software underpinning UK banking infrastructure are extremely dated. He said AI tools could compress weeks of manual code review into minutes. He added a note of caution, though, observing that these systems still generate false positives and require human oversight to validate their findings.

Anthropic initially restricted Mythos access to a group of roughly 42 organisations, the majority of them US technology companies, leaving European financial institutions largely on the outside.

OpenAI Moves to Fill the Access Gap

Former UK Chancellor George Osborne, now a senior executive at OpenAI, told the BBC that Bank of England Governor Bailey had not approached him directly before going public with his concerns. Osborne said his company had no interest in keeping GPT-5.5 Cyber restricted, but acknowledged it would not be universally available either. He framed the access question around ensuring that institutions defending democratic order hold these tools, while bad actors do not.

The nine UK institutions set to gain access include Lloyds Banking Group, HSBC, and Nationwide. NatWest and Santander already hold access through pre-existing agreements. OpenAI has also extended GPT-5.5 Cyber access to banks in Japan, Canada, and across the EU. Anthropic, by contrast, invested $100 million in subsidising Mythos previews, though that funding has so far bypassed UK finance entirely.

Read Next: AI Regulation Is Accelerating — Here Is What Markets Need to Know