U.K. Reviews Risks Linked To Anthropic AI Model

Regulators in the United Kingdom (U.K.) have begun discussions to assess risks linked to a new artificial intelligence (AI) model from Anthropic, according to a report by Financial Times.

Officials from the Bank of England, Financial Conduct Authority, and the Treasury are in talks with the National Cyber Security Centre to review risks to IT systems. The discussions follow findings linked to the model known as Claude Mythos Preview. Representatives from banks, insurers, and exchanges are expected to attend a briefing with regulators in the coming weeks to review cyber risk exposure. Authorities are examining how the model could affect system security and financial infrastructure.

The model is part of a controlled programme called Project Glasswing, under which selected organisations can test it for cyber defence use. Reports indicate it has identified vulnerabilities across software systems and browsers.