What happened
Microsoft temporarily banned employee use of Anthropic's Claude Fable 5, a Mythos-class AI model, citing new data retention policies. Anthropic's updated terms permit retaining prompts and outputs for 30 days, and content flagged by safety systems for up to two years, even for enterprise customers. This action prevents potential exposure of sensitive corporate and customer data. Concurrently, Microsoft terminated all Claude Code licenses effective June 30, directing employees to GitHub Copilot CLI, a move influenced by high operational costs and a strategic shift towards internal products.
Why it matters
Mandatory data retention by AI providers introduces significant operational security risks for enterprises. This necessitates security architects reassessing third-party AI model contracts, particularly regarding data handling and retention periods, to prevent inadvertent exposure of proprietary information. This follows Microsoft's prior internal adoption and subsequent termination of Claude Code licenses due to cost and strategic alignment. Procurement teams will need to scrutinise vendor terms for potential shifts that could force costly platform migrations or internal product mandates, impacting budget and workflow continuity.
