OpenAI o1 System Card
OpenAI Blog
OpenAI released safety documentation for its o1 and o1-mini models, describing red teaming and risk assessments conducted before launch. The evaluation followed OpenAI's Preparedness Framework, which systematizes frontier risk testing across multiple capability areas. The system card provides transparency on what safety testing was performed but does not indicate whether testing revealed significant risks or how findings influenced the final release decision.
Why it matters
This report outlines the safety work carried out prior to releasing OpenAI o1 and o1-mini, including external red teaming and frontier risk evaluations according to our Preparedness Framework.