Deep research System Card

In three linesOpenAI releases a System Card for "deep research" detailing safety work: external red teaming, frontier risk evaluations per Preparedness Framework, and mitigations for key risk areas prior to launch.

Summary generated by Claude — human-verified