Microsoft is restricting internal access to Anthropic's freshly released Claude Fable 5, The Verge reported, citing the AI startup's new data retention requirements as the reason. The block applies to the model picker that Microsoft employees use for internal versions of GitHub Copilot. All other Claude models remain available because they operate under Zero Data Retention (ZDR) rules.
The tension is ironic. Microsoft rolled out Claude Fable 5 to its GitHub Copilot and Foundry customers almost immediately after Anthropic launched the model on Tuesday. But internally, the company's legal teams are still evaluating whether the model is safe for employee use.
Anthropic released Claude Fable 5 as its first broad Mythos-class model, just weeks after telling the public the family was so capable at cybersecurity tasks that it was too dangerous to release. The company added prompt safeguards for the public version, and those safeguards come with a cost: data retention.
Claude Fable 5 requires Anthropic to retain prompts and outputs for 30 days to operate its new safety classifiers. Prompts flagged as violating usage policy can be stored for up to two years.
That's a non-starter for Microsoft's legal teams, who are concerned about customer data and confidential information flowing through a third party's retention pipeline.
Microsoft declined to comment on the situation, per The Verge. The restriction highlights a growing friction point in enterprise AI adoption. Companies that sell AI models to customers are also consumers of rival models, and data retention policies are creating legal headaches that standard API terms don't always address.
Microsoft's internal ban on Claude Fable 5 is a concrete example of how quickly those tensions surface when a model's safety architecture conflicts with a customer's data governance requirements.
