Anthropic launches Enterprise Frontier Safeguards

Anthropic launched Enterprise Frontier Safeguards, a tool that stores the data generated by its misuse-detection monitoring inside a customer’s own cloud account — Amazon S3, Azure Blob Storage or Google Cloud Storage — under the customer’s own encryption keys and access policies, rather than on Anthropic’s servers. Anthropic said the data would be reviewed by the customer’s own teams rather than by Anthropic staff.

The tool is intended to reconcile two things enterprise customers had been asking for separately: zero data retention, so that prompts and outputs are not kept by Anthropic after a request completes, and continued detection of misuse such as attempts to generate harmful content or circumvent safety controls, which normally requires retaining and inspecting some interaction data. Anthropic said it built the product with input from more than 100 enterprise customers across financial services, healthcare, manufacturing, telecoms, law, retail and the public sector, alongside Amazon Web Services, Google Cloud and Microsoft Azure.

The company said the rollout would happen in phases starting in autumn 2026, and that eligible customers would receive zero data retention on Fable 5 and Fable 5.1 in the meantime, ahead of Enterprise Frontier Safeguards becoming available. The product addresses a tension specific to running frontier models for regulated industries: those customers’ compliance requirements often prohibit a vendor from retaining their data at all, which otherwise leaves the vendor unable to see whether its own model is being misused inside that customer’s account.