On September 1, 2026, Anthropic announced a change to its data retention policy for business customers, introducing a new solution called Enterprise Frontier Safeguards.
This decision follows substantial feedback from enterprise users concerned about the previous policy, which mandated a 30-day retention of all traffic on its advanced AI models, Claude Fable 5 and Mythos 5, to mitigate misuse and cyber threats. Kate Jensen, Anthropic's head of Americas, indicated that the company invested considerable time collaborating with customers to develop this new solution.
The Enterprise Frontier Safeguards will allow businesses to manage their data more effectively, including automated safety monitoring without requiring human review from Anthropic. Notably, this service will be provided at no additional cost and will be compatible with both direct access to Anthropic's technology and through cloud providers.
Anthropic's previous data retention policy remains in effect for non-enterprise users. This strategic shift is crucial for Anthropic as it prepares for a potential IPO and seeks to differentiate itself from competitors like OpenAI, which recently launched its own data retention solution.
Additionally, Anthropic unveiled its latest models, Claude Fable 5.1 and Mythos 5.1, which promise enhanced capabilities in coding and research. The phased rollout of Enterprise Frontier Safeguards is expected to begin this fall, with Jensen expressing confidence in accelerated deployment, particularly for sensitive workloads