Deep research System Card
OpenAI published its safety playbook for AI research agents—here's what the red team found.
Why it matters
As AI agents move into knowledge work, safety governance and transparent risk disclosure are becoming competitive differentiators. OpenAI's system card signals the industry standard for pre-release safety validation.
The key facts
11 to knowOpenAI released system card for Deep Research product
Includes external red teaming results
Frontier risk evaluations per Preparedness Framework
Published Feb 25 2025
Addresses mitigations for key risk areas
Public safety governance documentation
External red teaming conducted prior to release
Frontier risk evaluations per OpenAI Preparedness Framework
Mitigations built into product for key risk areas
Published safety documentation on system-level risks
Indicates governance/safety-first release approach for research capabilities
Go to the source
OpenAI Blogopenai.com
Publisher excerpt: This report outlines the safety work carried out prior to releasing deep research including external red teaming, frontier risk evaluations according to our Preparedness Framework, and an overview of the mitigations we built in to address key risk areas.