Private Safety Processing keeps ZDR prompts out of OpenAI's hands
- Source
- x.com
- Date
Today we’re previewing Private Safety Processing, designed to let us keep offering Zero Data Retention while improving our safeguards. Even when benefiting from frontier intelligence, customers shouldn’t have to give up control of sensitive data. For ZDR deployments, content stays on infrastructure the customer controls. Automated systems look for patterns across related interactions and return limited safety signals, without exposing the underlying prompts or responses to OpenAI employees (even me!). We’re also developing an OpenAI-hosted option encrypted with customer-controlled keys. We’re testing this with early customers now and plan to begin rolling it out in September.
Context
OpenAI's Private Safety Processing preview proposed checking related AI interactions for safety risks while keeping their contents under customer control. The August 19, 2026 description said automated checks would run on customer-controlled infrastructure and return limited safety signals, without exposing the underlying prompts or responses to company employees.
The design addresses a tension in longer autonomous tasks: a risky pattern may span several interactions, while customers using Zero Data Retention want to retain control of sensitive content. The announcement also described a hosted option with customer-controlled encryption keys as under development. It presented a design being tested with early customers, with rollout planned for September.
Checking sign-in…
Loading comments…