UK regulators warn firms to limit risks from frontier AI models

The UK’s finance ministry, the Bank of England, and the Financial Conduct Authority issued a joint warning on Friday that firms must prepare for risks posed by frontier AI models. The regulators stated that these models already surpass the cyber capabilities of skilled practitioners, operating at higher speed, scale, and lower cost. If misused, they could amplify cyber threats to financial stability, market integrity, and customer safety. The statement emphasized that companies should not rely blindly on AI and must implement plans to mitigate potential risks. The warning follows concerns about Mythos, an AI product developed by Anthropic, which cyber experts have flagged as capable of escalating complex cyberattacks. Bank of England Governor Andrew Bailey previously highlighted Mythos as a major cybersecurity risk to the banking industry. The regulators noted that frontier AI models could challenge existing technological defenses, potentially exposing vulnerabilities in financial systems. Firms were urged to assess and address these risks proactively to safeguard operations and maintain trust in the financial sector. The joint statement underscores growing apprehensions about AI-driven cyber threats in an era where advanced models are increasingly integrated into critical infrastructure. Authorities stressed the need for vigilance and adaptive strategies to counter evolving risks in financial stability and cybersecurity.