OpenAI to allow third-party AI safety reviews in training — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- OpenAI plans to let third-party groups conduct technical safety evaluations of its AI models across the training, evaluation, and deployment phases.
- The third-party vetting would target safety risks during earlier phases of model development, not just deployment-stage testing.
Why it matters: By inviting third-party safety reviews across training, evaluation, and deployment phases rather than confining them to deployment-stage testing, OpenAI moves external auditing upstream into the proprietary model training process — expanding outside access across the full model-building pipeline.
Ask SkimNews
