Anthropic’s latest AI model, Claude Mythos, has triggered urgent cybersecurity risk assessments by UK financial regulators following concerns over its potential impact on critical systems.
Authorities including the Bank of England, Financial Conduct Authority, and the UK Treasury are reportedly working with the National Cyber Security Centre to evaluate risks associated with the model. Banks, insurers, and financial institutions are expected to receive briefings on potential threats in the coming weeks.
The concerns follow the launch of Claude Mythos Preview, part of Anthropic’s cybersecurity initiative Project Glasswing, which aims to detect vulnerabilities across software systems. The company claims the model has already identified thousands of high-severity and zero-day vulnerabilities, including long-standing flaws in major platforms.
Despite these capabilities, regulators are focusing on the dual-use risk of advanced AI, where tools designed for defense could potentially be exploited by malicious actors. Similar concerns have emerged in the United States, where senior officials, including Scott Bessent, have held discussions with major financial institutions such as JPMorgan Chase and Goldman Sachs to assess cyber risk exposure.
The scrutiny also comes amid ongoing tensions between Anthropic and U.S. authorities, including a legal dispute linked to national security concerns and supply chain risk classifications.
As AI adoption accelerates in cybersecurity and financial services, the situation highlights the growing need for robust governance, risk assessment, and regulatory oversight to balance innovation with security.
