Anthropic introduces customer-controlled safeguards for frontier AI

Anthropic has announced Enterprise Frontier Safeguards (EFS), a new enterprise offering designed to combine zero data retention (ZDR) with monitoring for serious AI misuse. The company detailed the plan in Anthropic News, saying EFS will begin rolling out in phases later this fall.

The system stores activity data in cloud infrastructure controlled by each customer rather than by Anthropic. Organizations can use their own encryption keys, access policies, audit logs, and storage accounts—including Amazon S3, Azure Blob Storage, or Google Cloud Storage. Automated systems analyze a rolling window of traffic for signals such as credential theft, offensive cyber activity, and attempts to develop biological capabilities. Alerts go to the customer, with no Anthropic employee required to conduct human review.

Anthropic says the design reflects work with more than 100 customers across finance, healthcare, manufacturing, telecom, law, retail, and government, alongside AWS, Google Cloud, and Microsoft Azure. Participants included companies such as Comcast, KPMG, Mastercard, Salesforce, and Visa, as well as members of the Analysis and Resilience Center for Systemic Risk.

EFS will support Claude Code, Claude Enterprise, the Claude Platform, Amazon Bedrock, Claude Platform on AWS, Google’s Agent Platform, and Microsoft Foundry. Eligible customers will receive ZDR on Fable 5 and Fable 5.1 until the new system is available.

The announcement addresses a central deployment problem: more capable, agentic models can require cross-session monitoring to detect sophisticated abuse, while regulated organizations may be unable to hand sensitive logs to an additional vendor. EFS aims to separate Anthropic’s detection capabilities from custody and review of the underlying data.

Source: Anthropic News