AI & agents
Who stores and reviews AI activity? Anthropic announces enterprise safeguards
- Published
- Source
- Anthropic announcement
- Source type
- Planned product
Summary
Anthropic announced Enterprise Frontier Safeguards (EFS). The proposed setup stores activity data in customer-controlled cloud infrastructure and sends automated detection findings to the customer’s reviewers. At announcement, a phased rollout was planned for the fall; the features were not immediately available to every customer.
Why it matters for our work
AI adoption discussions should cover where records remain and who can access them, as well as whether data is used for training. We also need to distinguish retention terms for external AI connections from those for internal work records. A provider’s policy alone does not establish retention conditions across our entire system.
Translated from the Korean original. Summaries may be translated and edited. Commentary reflects our perspective; forecasts remain the source’s views.