Anthropic has announced Enterprise Frontier Safeguards (EFS), a new solution designed to combine the privacy of zero data retention (ZDR) with enhanced safeguards for detecting misuse of advanced AI models.
EFS stores data in cloud infrastructure controlled by the customer rather than Anthropic, enabling enterprises to retain greater control over their data while supporting monitoring capabilities required to identify potentially harmful activity. The solution will roll out to customers in phases from later this autumn.
Eligible customers will receive ZDR for Claude Fable 5 and Fable 5.1 until EFS becomes available.
Anthropic said EFS was developed in collaboration with more than 100 customers across financial services, healthcare, manufacturing, telecommunications, law, retail and the public sector. The company also worked with cloud providers Amazon Web Services, Google Cloud and Microsoft Azure.
The solution will be supported across Claude Code, Claude Enterprise, the Claude Platform, Amazon Bedrock, Claude Platform on AWS, Google’s Agent Platform and Microsoft Foundry.
According to Anthropic, the increasing intelligence and agentic capabilities of advanced models create new risks, including fraud, sophisticated cyberattacks and autonomous destructive behaviour. The company said some attempted misuse has involved the theft or misappropriation of enterprise credentials.
Anthropic noted that sophisticated misuse can span multiple tasks, sessions and accounts, making it difficult to detect through analysis of individual interactions alone. Monitoring and correlating activity over time can therefore be important for identifying abnormal behaviour.
The company introduced 30-day data retention with Fable 5 to enable such monitoring, while emphasising that the policy was not introduced to train models on enterprise data. Anthropic said it has never trained on enterprise data without explicit permission.
EFS was subsequently designed to address concerns among regulated enterprises that data retention could conflict with privacy and compliance requirements. Anthropic worked with security, product, compliance and delivery teams, including the Analysis and Resilience Center for Systemic Risk (ARC), as well as organisations such as Comcast, KPMG, Mastercard, Salesforce and Visa.
