SafetyPage

Lab system cards and frameworks · System card · OpenAI

Deep Research System Card

25 Feb 2025

Read at openai.com ↗Read at sourceAll rights reserved

Outline · 36 pages

  1. Introductionp. 1
  2. Model data and trainingp. 1
  3. Risk identification, assessment and mitigationp. 2
  4. External red teaming methodologyp. 2
  5. Evaluation methodologyp. 2
  6. Observed safety challenges, evaluations and mitigationsp. 2
  7. Prompt Injectionp. 3
  8. Disallowed Contentp. 6
  9. Privacyp. 8
  10. Ability to run codep. 8
  11. Biasp. 9
  12. Hallucinationp. 9
  13. Preparedness Framework Evaluationsp. 10
  14. Addressing browsing-based contaminationp. 11
  15. Preparedness Mitigationsp. 12
  16. Cybersecurityp. 13
  17. Chemical and Biological Threat Creationp. 17
  18. Radiological and Nuclear Threat Creationp. 22
  19. Persuasionp. 24
  20. Model Autonomyp. 27
  21. Conclusion and Next Stepsp. 34
  22. Acknowledgmentsp. 35