Anthropic has introduced a mandatory 30-day data retention and review policy for its newest AI model, Claude Fable 5, and other Mythos-class models, requiring all user data—prompts and outputs—to be collected without exception, including for enterprise partners.
- Claude Fable 5 user data is collected and retained for 30 days.
- New policy overrides zero data retention agreements for enterprises.
- Model designed with safety guardrails but data collection is unavoidable.
What happened
Anthropic publicly released Claude Fable 5, a ‘safe for general use’ version of its powerful Mythos AI model, along with updated data retention policies. The company now retains all prompts submitted and outputs generated by Mythos-class models, including Fable 5, for 30 days for trust and safety purposes.
This change applies universally, with no exceptions allowed even for enterprise customers or API partners. Prior zero data retention agreements that applied to other Claude models no longer extend to Fable 5 or any Mythos-class models, marking a significant shift in Anthropic’s handling of user data.
Why it matters
The mandatory data retention policy impacts organizations relying on Anthropic’s AI models for sensitive or proprietary tasks, as their inputs and outputs will be stored and reviewed. This can have implications for privacy, compliance, and intellectual property protection, especially for enterprise clients that previously benefited from zero data retention agreements.
While Anthropic emphasizes the safety and powerful capabilities of these models, enterprises must now reassess their risk tolerance and data governance strategies when integrating Fable 5 or any Mythos-based AI services, given that data cannot be excluded from collection and review.
What to watch next
Clients and partners of Anthropic should closely monitor any further updates or clarifications on data usage policies, especially regarding how data is stored, protected, and potentially anonymized during the 30-day retention window. Legal and compliance teams will need to review existing contracts in light of this new policy change.
The wider AI industry may also respond to Anthropic’s approach to data retention as regulatory scrutiny around AI safety and privacy increases. Enterprises could seek alternative providers offering stronger data privacy guarantees if this policy is deemed incompatible with their operational or regulatory requirements.