SafetyPage

Lab system cards and frameworks · System card · OpenAI

Addendum to GPT-5.2 System Card: GPT-5.2-Codex

18 Dec 2025

Read at openai.com ↗Read at sourceAll rights reserved

Outline · 22 pages

  1. Introductionp. 4
  2. Baseline Model Safety Evaluationsp. 4
  3. Disallowed Content Evaluationsp. 4
  4. Jailbreaksp. 5
  5. Product-Specific Risk Mitigationsp. 6
  6. Agent Sandboxp. 6
  7. Network Accessp. 6
  8. Model-Specific Risk Mitigationsp. 7
  9. Cyber Safetyp. 7
  10. Risk Descriptionp. 7
  11. Mitigation: Safety Trainingp. 7
  12. Avoid Data-Destructive Actionsp. 8
  13. Risk Descriptionp. 8
  14. Mitigation: Safety Trainingp. 8
  15. Preparednessp. 8
  16. Capabilities Assessmentp. 9
  17. Biological and Chemicalp. 9
  18. Multimodal Troubleshooting Virologyp. 9
  19. ProtocolQA Open-Endedp. 9
  20. Tacit Knowledge and Troubleshootingp. 10
  21. Troubleshooting Benchp. 11
  22. Cybersecurityp. 11
  23. Capture-the-Flag (Professional)p. 13
  24. CVE-Benchp. 14
  25. Cyber Rangep. 15
  26. External Evaluations by Irregularp. 16
  27. AI Self-Improvementp. 17
  28. OpenAI PRsp. 18
  29. MLE-bench-30 (n=30)p. 18
  30. Paperbench-10 (n=10)p. 20
  31. OpenAI-Proof Q&Ap. 21