OpenAI has reaffirmed its commitment to Zero Data Retention (ZDR) for eligible API customers and is previewing a new system called Private Safety Processing. This initiative aims to strengthen safeguards for advanced AI while preserving data privacy. Private Safety Processing is designed to identify patterns across related interactions without granting OpenAI personnel access to the underlying content.
Key Points
- OpenAI does not retain prompts or model responses for eligible API customers after a request is processed under Zero Data Retention.
- Customer content is not available to OpenAI personnel for review, and enterprise customer data is not used for model training unless customers explicitly opt-in.
- Private Safety Processing is designed to identify patterns across multiple related interactions, addressing risks that may not be visible in single interactions.
- This new system extends existing automated protections in ZDR deployments, which evaluate interactions individually.
- For ZDR deployments, customer content remains on customer-controlled infrastructure, or it can be stored on OpenAI infrastructure, encrypted with customer-controlled keys.
- OpenAI plans to begin rolling out Private Safety Processing and release a technical white paper in September.
- Images flagged for potential Child Sexual Abuse Material (CSAM) will continue to be retained for manual review and reporting, even in Zero Data Retention deployments, as required by law.
Context
According to OpenAI, as AI models undertake longer and more complex tasks, potential risks may only become apparent when viewed across multiple interactions. Existing ZDR-compatible safety systems evaluate each interaction in isolation. Private Safety Processing aims to address this by identifying patterns across related interactions without exposing the content to OpenAI personnel. This approach is intended to allow for more comprehensive safety monitoring while upholding privacy commitments.
Why It Matters
This development addresses a critical tension for builders and enterprises: the need for advanced AI safety monitoring that does not compromise data privacy or conflict with security obligations. By enabling pattern identification across interactions without direct content access, OpenAI aims to provide a solution that supports the adoption of more capable AI systems in sensitive environments.
What To Do
- Note that Private Safety Processing is designed to be compatible with Zero Data Retention commitments.
- Watch for the planned rollout of Private Safety Processing and the release of a technical white paper in September.
- Review the implications for existing ZDR commitments as more updates are shared by OpenAI.
- Consider how this approach might impact the use of frontier models in applications handling sensitive data.