Developing Enterprise Frontier Safeguards with our customers

Anthropic announced Enterprise Frontier Safeguards (EFS), a solution combining the privacy of zero data retention (ZDR) with state-of-the-art safeguards for detecting misuse. The move addresses the dilemma posed by 'mythos-class' models like Claude Fable 5.1, whose increased intelligence and agentic capabilities bring both potential for misuse and autonomous misbehavior. Anthropic cites substantial evidence of attempted misuse, from fraud to sophisticated cyberattacks involving agents, including theft of enterprise credentials that is hard to detect without monitoring traffic and abnormal behavior.

Because sophisticated misuse can span multiple sessions and accounts, EFS stores data in cloud infrastructure controlled by the customer, not Anthropic, allowing correlation across time and accounts while preserving ZDR-like privacy. This departs from Anthropic's 30-day data retention policy introduced with Fable 5; the company emphasizes it has never trained on enterprise data without permission and never will. EFS was developed in close collaboration with over 100 customers across financial services, healthcare, manufacturing, telecom, law, retail, and the public sector, plus cloud partners AWS, Google Cloud, and Microsoft Azure. It will be supported on Claude Code, Claude Enterprise, the Claude Platform, Amazon Bedrock, Claude Platform on AWS, Google's Agent Platform, and Microsoft Foundry.

The solution was designed with feedback from security, product, compliance, and delivery teams, including the Analysis and Resilience Center for Systemic Risk (ARC), whose members include CISOs of major US banks such as Goldman Sachs, Morgan Stanley, Citi, Bank of America, and Wells Fargo. EFS will roll out to customers in phases starting later this fall, and to smooth the transition, eligible customers will receive ZDR on Fable 5 and Fable 5.1 until EFS is ready.