Safety, alignment & governance
Offering Zero Data Retention for frontier models
Private Safety Processing aims to detect misuse patterns across related interactions while keeping underlying prompts and responses inaccessible to OpenAI personnel.
Why it matters: Organizations handling sensitive records or research may need broader misuse detection without allowing a provider to retain their content for monitoring. This approach is intended to address that tension while keeping customer investigation and optional disclosure in the customers’ hands.