SafetyPage

Lab system cards and frameworks · System card · OpenAI

Update to GPT-5 System Card: GPT-5.2

11 Dec 2025

Read at openai.com ↗Read at sourceAll rights reserved

Outline · 27 pages

  1. Introductionp. 4
  2. Model Data and Trainingp. 4
  3. Baseline Model Safety Evaluationsp. 4
  4. Disallowed Content Evaluationsp. 4
  5. Jailbreaksp. 5
  6. Prompt Injectionp. 6
  7. Visionp. 6
  8. Hallucinationsp. 7
  9. Healthp. 9
  10. Deceptionp. 9
  11. Cyber Safetyp. 10
  12. Multilingual Performancep. 11
  13. Biasp. 11
  14. Preparedness Frameworkp. 12
  15. Capabilities Assessmentp. 12
  16. Biological and Chemicalp. 13
  17. Multimodal Troubleshooting Virologyp. 13
  18. ProtocolQA Open-Endedp. 14
  19. Tacit Knowledge and Troubleshootingp. 14
  20. TroubleshootingBenchp. 15
  21. Cybersecurityp. 16
  22. Capture the Flag (CTF) Challengesp. 17
  23. CVE-Benchp. 19
  24. Cyber Rangep. 20
  25. External Evaluations for Cyber Capabilitiesp. 21
  26. AI Self-Improvementp. 21
  27. OpenAI PRsp. 22
  28. MLE-Benchp. 23
  29. PaperBenchp. 24
  30. OPQAp. 25
  31. Research Category Update: Sandbaggingp. 26