Anthropic has announced the launch of Enterprise Frontier Safeguards (EFS), a new solution designed to address the security and privacy concerns of enterprise customers using frontier AI models. EFS combines zero data retention (ZDR) with state-of-the-art safeguards for detecting misuse, while allowing customers to store data in their own cloud infrastructure rather than on Anthropic's servers.
The development of EFS was a collaborative effort involving more than 100 customers across industries such as financial services, healthcare, manufacturing, telecom, law, retail, and the public sector, as well as cloud partners including Amazon Web Services, Google Cloud, and Microsoft Azure. The solution will be supported on platforms like Claude Code, Claude Enterprise, Amazon Bedrock, and Microsoft Foundry.
EFS addresses the challenge of detecting sophisticated misuse that spans multiple sessions and accounts. Traditional methods that analyze interactions separately and discard data immediately are insufficient. EFS enables automated monitoring over a rolling window of traffic, with flags sent directly to customers for review, eliminating the need for Anthropic human review.
Customers have full control over data storage and management, including the ability to use their own encryption keys and access policies. Activity data can be stored in the customer's cloud account (e.g., Amazon S3, Azure Blob Storage, or Google Cloud Storage). These controls are opt-in and do not affect model behavior, API pricing, or rate limits. Anthropic does not charge for EFS; customers only pay their cloud provider for storage and related fees.
EFS will roll out in phases starting later this fall. Eligible customers will receive ZDR on Fable 5 and Fable 5.1 until EFS is fully available. Anthropic also announced improvements in alignment and security, including a research preview of the Model Hardware Standard (MHS) and expanded support for scientists with free Claude access for 10,000 researchers.