Published collection
OpenAI slows model scaling due to critical cyber capability threshold
This measure shows that the pace of frontier model training is being constrained by cybersecurity capability evaluations.
- Published entries
- 2
- Sources
- 2
- Date range
- 2026-08-18 to 2026-08-19
- Primary labels
- Industry · 行业 · Policy / safety / copyright · OpenAI 前沿 RL 训练 · OpenAI · Hugging Face
Published evidence
Every entry keeps its summary and a path back to the source context.
OpenAI slows model scaling due to critical cyber capability threshold
Due to the OpenAI-Hugging Face incident and the possibility that the Astra model may have reached a critical cybersecurity capability threshold, OpenAI temporarily slowed model scaling, paused reinforcement learning training for its latest deployed model for two weeks, and put its largest frontier RL run on hold.
Why it mattersThis measure shows that the pace of frontier model training is being constrained by cybersecurity capability evaluations.
Original sourceOpenAI pauses RL training for latest model to harden research environment
OpenAI says it temporarily paused RL training for latest models used in deployment for two weeks to harden and red-team the research environment and expand monitoring coverage; its largest frontier RL run remains paused.
Why it mattersThis will affect the pace of frontier model training and highlights the need for internal training safety safeguards.
Original source