OpenAI has announced Private Safety Processing (PSP), a system that allows the company to detect safety risks in AI interactions without actually storing or reviewing customer data. The system transmits only limited safety signals while keeping the content of exchanges completely opaque to OpenAI, maintaining Zero Data Retention commitments for API and enterprise clients. Businesses can either keep their data entirely within their own infrastructure or store it encrypted on OpenAI’s servers with encryption keys they control. The feature was announced on August 19, 2026, with a phased rollout to API customers beginning on September 22, 2026. It remains exclusive to enterprise and API offerings, unavailable for consumer ChatGPT tiers including Free, Plus, Go and Pro plans.
Source: Read the original article
