Anthropic Says Mythos AI Model Available to UK Banks in 'Next Week'

Summary

Pip White, Anthropic's head for UK, Ireland, and northern Europe, announces that the Mythos AI model will be released to UK financial institutions in the coming week. The model, which excels at identifying severe cybersecurity vulnerabilities, has been initially provided to a small cohort including Microsoft and some US banks. Anthropic emphasizes a controlled and safety-focused release strategy to manage risks, prioritizing responsible AI development over commercial considerations.

• Anthropic's Mythos AI model is planned for release to UK financial institutions in the next week.
• Mythos has been tested by a small cohort including Microsoft and select US banks.
• The model demonstrates strong capabilities in finding cybersecurity vulnerabilities in operating systems and web browsers.
• Anthropic adopted a controlled release strategy to learn from initial users in a contained manner.
• UK financial institutions have shown significant interest and engagement with Anthropic regarding Mythos access.
• The company prioritizes safe and responsible AI development, even if it slows commercial release.
• Anthropic plans to maintain safeguards and intentional release processes for future models.
• The interview highlights the balance between AI innovation and security risks in highly regulated sectors.