Offering Zero Data Retention for frontier models

The system is being tested with early customers, with rollout and a technical white paper planned for September.

Automated systems analyze patterns across related interactions without giving OpenAI personnel access to prompts or responses.
Customer content can remain on customer-controlled infrastructure or use OpenAI storage encrypted with customer-controlled keys.
When potential misuse is detected, OpenAI receives only a limited signal identifying the activity type for possible enforcement.
Customers may use their own records to investigate decisions and voluntarily share relevant information for appeals or verified-abuse investigations.
Images flagged as potential CSAM remain subject to retention, manual review, and legally required reporting, including in ZDR deployments.

Eligible API customers can retain control of sensitive content while gaining safety monitoring across multi-interaction and agentic tasks.

Sources