Copilot's Fable 5.1 turns data retention on by default

Turning on Claude Fable 5.1 in GitHub Copilot comes with a data-handling condition the other Claude models there do not have. According to GitHub's changelog, Anthropic retains prompts and outputs from Fable 5.1 to run safety classifiers that detect harmful use. GitHub says that retained data is not used to train Anthropic's models, and that every other Claude model in Copilot except Fable 5 and Fable 5.1 continues to run under zero data retention.

The model is offered on the Pro+, Max, Business and Enterprise plans. For Business and Enterprise organizations, an administrator has to switch on a Fable 5.1 policy that starts off, and enabling it counts as acknowledging the retention requirement; leaving it off keeps the model unavailable to the organization.

Eligible enterprises can instead use Fable 5.1 and Fable 5 with zero retention through the end of the calendar year, under a time-bound exemption while Anthropic rolls out Enterprise Frontier Safeguards. After that, GitHub says continued use will require those safeguards, which add automated safety monitoring along with more choices about where data is kept, including storage and encryption keys under the customer's control. Eligibility goes through the GitHub account team, not Support, and approval does not switch anything on: an administrator still has to enable the policy, at which point requests are routed to zero-retention endpoints.