The company's new 'Private Intelligence' suite will allow corporate clients to run encrypted AI queries through trusted execution environments to ensure data is isolated during processing. Debuted at DevDay, the toolkit includes a preview of 'Private Inference' arriving this fall and a 'Zero Data Retention' feature that permits automated safety reviews without granting OpenAI staff access to the actual contents of client prompts.

This rollout aligns OpenAI with rivals including Google, Meta, and Apple, all of whom have integrated confidential computing to protect sensitive user information. By implementing strict, verifiable controls over how data is handled, the firm aims to establish confidential computing as the standard for frontier AI infrastructure and increase the confidence of enterprise users.