Anthropic Introduces Enterprise Frontier Safeguards for AI Safety and Privacy

Anthropic has launched Enterprise Frontier Safeguards (EFS), a comprehensive safety solution for enterprise AI applications. EFS combines the privacy benefits of zero data retention (ZDR) with state-of-the-art safeguards for detecting misuse, allowing data to be stored securely in the customer's cloud infrastructure rather than Anthropic's. The rollout will begin later this fall, with eligible customers receiving ZDR on Fable 5 and Fable 5.1 during the transition.

Developed in collaboration with over 100 customers across industries such as finance, healthcare, manufacturing, telecom, law, retail, and the public sector, EFS aims to address the challenges of monitoring AI misuse while respecting enterprise data privacy. The solution supports cloud platforms including Amazon Web Services, Google Cloud, and Microsoft Azure, and will be compatible with models like Claude Code, Claude Enterprise, and Claude Platform.

The need for EFS stems from the increasing sophistication of AI misuse, including fraud, cyberattacks, and autonomous destructive behaviors. Detecting such misuse requires monitoring traffic over time and across accounts, which is difficult without data retention. To this end, Anthropic introduced a 30-day data retention policy starting with Fable 5, solely for safety monitoring purposes, explicitly not for training.

Feedback from security, compliance, and product teams informed the design of EFS, ensuring that data control remains with the enterprise. Customers can store activity data in their own cloud environments with their encryption and access policies, facilitating compliance with industry regulations. Automated monitoring signals are sent directly to the enterprise for review, enabling human oversight by authorized personnel.

Anthropic emphasizes that the solution is built with input from industry leaders, including the chief information security officers of major US banks and companies like Mastercard and Salesforce. The design aims to meet regulatory standards for insider risk monitoring and misuse detection, providing enterprises with a secure, controlled environment to deploy advanced AI models safely.