OpenAI reaffirms its Zero Data Retention (ZDR) commitment for eligible API customers, meaning prompts and responses are not retained after processing and are not used for training without opt-in. It previews a new feature, Private Safety Processing, which will detect misuse patterns across multiple related interactions without giving OpenAI staff access to underlying content. Content stays on customer-controlled infrastructure or is encrypted with customer-controlled keys when stored by OpenAI, with only narrow risk signals surfaced to OpenAI. Testing is underway with early customers, and a broader rollout plus a technical white paper is planned for September.
Table of contents
Why safety systems need to evolveHow Private Safety Processing worksPrivacy and safety built with and for our customersQuestions this post answers
What is Private Safety Processing from OpenAI and how does it relate to Zero Data Retention?
Private Safety Processing is a new OpenAI capability, previewed for eligible API customers, that detects misuse patterns across multiple related interactions without giving OpenAI staff access to the underlying content. It builds on existing Zero Data Retention (ZDR) safety systems that evaluate interactions individually, extending pattern detection across sessions while only surfacing a narrow signal about the type of activity involved, not the content itself. Track how vendor data-handling changes like this affect enterprise AI adoption decisions on daily.dev.
Does OpenAI use customer API data to train its models under Zero Data Retention?
No, under Zero Data Retention OpenAI does not retain prompts or responses after a request is processed, and enterprise customer data is not used to train models unless the customer explicitly opts in. Customer content also is not available to OpenAI personnel for review, except images flagged for potential CSAM, which are retained for legally required manual review and reporting. Compare AI vendor data-retention and training policies before committing on daily.dev.
When is OpenAI rolling out Private Safety Processing and what will customers get?
OpenAI plans to begin rolling out Private Safety Processing and publish a technical white paper in September, following testing with early customers. Content will remain on customer-controlled infrastructure for ZDR deployments, or be stored on OpenAI infrastructure encrypted with customer-controlled keys, with OpenAI receiving only limited safety signals rather than access to the content. Follow enterprise AI safety and privacy rollouts like this on daily.dev before you plan integrations.