Anthropic released Claude Fable, its first Mythos-class AI model, yesterday and it is already causing concerns inside Microsoft. Sources indicate that Microsoft is limiting the use of Claude Fable 5 for employees because of Anthropic's new data retention requirements.
Internal Access Restricted
While Microsoft quickly rolled out Claude Fable 5 to its GitHub Copilot and Foundry customers, the model is reportedly not available in the model picker that Microsoft employees use for internal versions of GitHub Copilot. All other Claude models remain available internally at Microsoft, as they operate under Zero Data Retention (ZDR) rules.
Legal Evaluation Underway
Microsoft has been informing employees that its legal teams are evaluating changes to Anthropic's data retention requirements. The primary concerns revolve around customer data and confidential information. It is not yet clear whether Microsoft's legal teams will approve Claude Fable for internal use.
Claude Fable 5 requires data retention to operate Anthropic's new safety classifiers, meaning Anthropic retains prompts and outputs and deletes this data after 30 days. Some prompts and outputs can even be stored for up to two years if they are flagged as violating Anthropic's usage policy, raising legal concerns about how Microsoft should use Claude Fable 5.
Microsoft declined to comment on the matter.
Background on Claude Fable 5
Claude Fable 5 is the first broad release from Anthropic's Mythos class of AI models. It arrives just weeks after the company stated that the family was so capable at cybersecurity tasks that it was too dangerous to release publicly. Anthropic has implemented prompt safeguards to make Fable 5 less dangerous, leading to these challenging data retention changes.
Update, June 10th: Added that Microsoft declined to comment.
