OpenAI unveils Private Intelligence to enhance AI data privacy

OpenAI has rolled out what it calls Private Safety Processing, a system designed to solve one of enterprise AI’s most persistent tensions: how do you monitor for safety risks across complex AI interactions without actually looking at the data? How Private Safety Processing actually works PSP is built to maintain OpenAI’s Zero Data Retention commitments for API and enterprise clients. That means no customer data gets stored, reviewed, or used for training. The system transmits only limited safety signals to OpenAI, keeping the actual content of interactions completely opaque to the company. As AI deployments have grown more sophisticated, they increasingly involve longer sequences of interactions rather than simple one-off queries. A single exchange might look perfectly benign. String twenty together and you might see someone systematically probing safeguards or an AI agent drifting off-mission. PSP is designed to catch those multi-turn patterns without compromising the privacy guarantees that enterprise customers demand. Customers get two options for their data: keep it entirely within their own infrastructure, or store it encrypted on OpenAI’s servers with encryption keys that the...