OpenAI previewed Private Safety Processing while Anthropic now logs 30 days.

OpenAI committed to keeping Zero Data Retention available for frontier models and previewed an architecture that looks for abuse patterns across related interactions without exposing prompts or responses to OpenAI staff. Customer data can stay on customer infrastructure or sit with OpenAI under customer-held encryption keys, with rollout planned for September. Axios framed it against Anthropic, which retains prompts and outputs for 30 days including for orgs that previously had ZDR, on the argument that the window is what catches multi-request attacks. (Axios) Both positions are defensible and they're genuinely incompatible. If you're picking a provider for regulated data, this is now a real differentiator rather than a checkbox.