Microsoft has begun restricting employees from using Anthropic's Claude Fable 5, its first Mythos-class AI model, due to new data retention policies imposed by Anthropic. According to sources cited by The Verge, the model is unavailable in the internal model picker for GitHub Copilot, though all other Claude models remain accessible under existing Zero Data Retention (ZDR) rules.
The restriction stems from Anthropic's updated 30-day data retention requirements, which conflict with Microsoft's internal data handling policies. The Verge reports that Microsoft's legal teams are actively evaluating the changes, signaling deeper compliance concerns within the technology giant.
Notably, Microsoft has already rolled out Claude Fable 5 to its GitHub Copilot and Foundry customer base. This split approach means external clients can access the new model, while internal employees cannot, highlighting a tension between product offerings and corporate data governance.
The move underscores a growing friction point in enterprise AI adoption: balancing model access with data privacy. As competitors deepen AI integrations, such data retention disputes may become more common, shaping how companies like Microsoft negotiate terms with model providers like Anthropic.
Anthropic's shift from ZDR to a retention period suggests a strategic pivot toward data collection, potentially enhancing model training. Critics argue this may erode trust among security-conscious enterprise clients who prioritize data sovereignty.