OpenAI has offered nine major UK banks access to its cyber security AI tool, GPT-5.5 Cyber, after rival Anthropic blocked them from using its tool, Claude Mythos. Both AI systems are designed to identify hidden vulnerabilities in digital infrastructures and have been noted for their performance, as confirmed by the AI Security Institute, which found both tools exhibited similar effectiveness. This situation comes amid growing concerns from finance ministers and central bankers about the potential risks advanced AI cyber tools might pose to the integrity of financial systems. Bank of England governor Andrew Bailey recently expressed frustration that UK banks could not access Mythos for securing their digital systems, highlighting the competitive landscape between OpenAI and Anthropic in the cyber security domain.

OpenAI: OpenAI develops advanced AI models and applications focused on various domains including security. In this news, the company has extended an offer to nine major UK banks for access to its GPT-5.5 Cyber tool after their rival blocked previews. Senior executive George Osborne, formerly UK Chancellor, publicly discussed the company’s approach to sharing the tool selectively.
UK banks: Major UK banks operate extensive digital systems that require ongoing security testing and maintenance. They have been denied preview access to Anthropic’s Mythos tool for evaluating their own infrastructure. This limitation has led to an alternative offer from OpenAI for a comparable cyber security AI solution.
Anthropic: Anthropic builds AI systems including the Claude family of models with applications in specialized tasks. It has restricted UK banks from preview access to its Claude Mythos cyber security tool. The decision has drawn regulatory attention and prompted competing offers from other providers.
Andrew Bailey: Andrew Bailey serves as Governor of the Bank of England with oversight of financial stability and banking practices. He publicly noted last week that UK banks still lack access to Mythos for testing their digital systems and applications. His comments highlight ongoing barriers for regulated institutions seeking advanced AI security tools.
George Osborne: George Osborne is a former UK Chancellor now working as a senior executive at OpenAI. He addressed the situation in media comments, stating that Bailey had not contacted him directly and confirming OpenAI’s intent to make its tool available under controlled conditions similar to Mythos.

`json
{
“Testing”: “The AI Security Institute evaluated both Mythos and GPT-5.5 Cyber and found they achieved comparable performance levels on assigned tasks.”,
“Regulation”: “Finance ministers, central bankers, and financiers have raised concerns that advanced AI cyber tools could pose risks to the security of financial systems.”,
“Competition”: “OpenAI and Anthropic are direct rivals developing specialized AI tools for identifying security weaknesses in digital systems.”
}
`