OpenAI Previews Private Safety Processing to Keep Zero Data Retention While Anthropic Now Logs 30 Days

Axios·high signalXBlueskyLinkedInCopy link

OpenAI committed on August 19 to keeping Zero Data Retention available for frontier models and previewed Private Safety Processing, an architecture that looks for abuse patterns across related interactions without exposing prompts or responses to OpenAI staff. Customer data can stay on customer-controlled infrastructure or sit with OpenAI under customer-held encryption keys; early customers are testing now with rollout planned for September. Axios framed it as a direct contrast with Anthropic, which retains prompts and outputs for 30 days, including for organizations that previously had ZDR, arguing that window is what catches multi-request attacks.