SafetyPage

Lab system cards and frameworks · System card · OpenAI

Addendum to GPT-4o System Card: Native image generation

Read at openai.com ↗Read at sourceAll rights reserved

Outline · 13 pages

  1. Introductionp. 1
  2. Observed Safety Challenges, Evaluations, and Mitigationsp. 1
  3. Safety Challenges: New risks from native image generationp. 1
  4. Safety stackp. 2
  5. Evaluationsp. 2
  6. External, manual red teamingp. 3
  7. Automated red teamingp. 3
  8. Offline testing using real-world scenariosp. 4
  9. Discussion of specific risk areasp. 4
  10. Child Safetyp. 4
  11. Artist Stylesp. 5
  12. Public Figuresp. 6
  13. Biasp. 6
  14. Other risk areas evaluatedp. 9
  15. Our approach to provenancep. 11
  16. Conclusionp. 11