FRONTIER · US
Federal Reserve lacked access to Anthropic's latest model amid banking cybersecurity push
The Fed flagged security risks in Anthropic's Claude Mythos model but did not have access to it as of mid-July, while other institutions were patching vulnerabilities. Underscores coordination gaps in AI safety disclosure to regulators.
WHY IT MATTERS
Central banks need real-time access to frontier models used by banks to assess systemic risk; lag in access hampers threat assessment and policy response timing.
Source: CNBC · 2026-07-22