Microsoft Restricts Claude Fable 5 Access Amid AI Safety Review

Microsoft restricted internal employee access to Anthropic’s Claude Fable 5 because the model requires storing prompts and outputs for at least 30 days (and up to 2 years for flagged content) as part of Anthropic’s safety monitoring. The article frames this as a governance and compliance concern for enterprises that handle proprietary or regulated data, notes Microsoft’s broader pattern of limiting use of certain external AI tools (including moving developers from Claude to GitHub Copilot), and highlights the tension between safety oversight and corporate data-retention/privacy policies.