Microsoft restricts employee use of Anthropic's Claude Fable over data retention concerns

Jun 10, 2026

Microsoft is restricting internal employee use of Anthropic's new Claude Fable 5 AI model over data retention concerns. The policy, required for new safety classifiers, retains user prompts and outputs for 30 days. Microsoft's legal team is evaluating the changes due to concerns about confidential data, though the model remains available to external customers on platforms like GitHub Copilot.

Microsoft employee usage restrictions

▪Other Anthropic Claude models remain available for Microsoft employees because they operate under Zero Data Retention (ZDR) rules.
▪Claude Fable 5 is not available in the internal model picker for Microsoft employees, but it is available to external customers via GitHub Copilot and Foundry.
▪Microsoft's legal teams are evaluating Anthropic's data retention changes, citing concerns about customer data and confidential information.
▪Microsoft is limiting its employees' use of Anthropic's Claude Fable 5 AI model due to new data retention requirements.

Anthropic data retention policy

▪The data retention policy for Claude Fable 5 is required to operate Anthropic's new safety classifiers.
▪Anthropic's policy for Mythos-class models like Claude Fable 5 retains prompts and outputs for 30 days for trust and safety purposes.
▪Prompts and outputs flagged as violating Anthropic's usage policy can be retained for up to two years.

Claude Fable 5 launch

▪In late May 2026, Anthropic raised $65 billion at a post-money valuation of $965 billion, placing it ahead of rival OpenAI.
▪In early June 2026, Anthropic confidentially filed for a U.S. initial public offering, without disclosing the size or terms.
▪Anthropic released Claude Fable 5, a public version of its Mythos AI model, on June 9, 2026.
▪Claude Fable 5 was released with safeguards to prevent its use in high-risk areas such as cybersecurity.

Claude Fable 5 capabilities

▪Claude Fable 5 is the most powerful model Anthropic has made available for wider use, with noted performance in software engineering and analytics.
▪Anthropic previously considered the Mythos model family too dangerous for public release due to its advanced cybersecurity capabilities.