Microsoft limits employee use of Anthropic's Claude Fable 5 over data retention concerns

Microsoft has begun limiting its employees' access to Anthropic's advanced AI model, Claude Fable 5, due to concerns over the artificial intelligence startup's new data retention requirements. The Verge reported Wednesday that sources familiar with the matter revealed the restrictions, which are currently being evaluated by Microsoft's legal teams.

Anthropic announced the rollout of Claude Fable 5, a public version of its powerful Mythos AI model, on Tuesday. The company has implemented guardrails to prevent its use in sensitive areas like cybersecurity. Anthropic highlights the model's strong performance in software engineering and analytics, positioning it as their most capable model available for broad use.

The core of Microsoft's concern lies in Anthropic's updated data retention policy. Under this policy, prompts submitted to and outputs generated by Mythos-class models are kept for 30 days for trust and safety monitoring. If flagged by safety classifiers as violating usage policies, this data can be retained for up to two years.

It remains unclear whether Microsoft's legal departments will ultimately approve Claude Fable 5 for internal use. The company's legal teams are specifically assessing the implications of these retention changes on customer data and confidential information. Neither Microsoft nor Anthropic immediately responded to requests for comment.