OpenAI open to slowing AI development as safety concerns grow
Key Points
- Altman reportedly told staff OpenAI could pace advanced AI
- OpenAI has already temporarily slowed selected frontier development
- Industry-wide pacing would require coordination among competing
OpenAI CEO Sam Altman reportedly told employees at a company-wide meeting this week that the company is open to slowing the development of its most advanced AI systems, Bloomberg reported.
Altman said OpenAI could pace development alongside other AI labs, although some competitors may not agree, according to the report. The comments do not amount to an announced company-wide slowdown.
OpenAI has already slowed frontier development
OpenAI has previously put that principle into practice on a limited basis. In August, the company temporarily slowed selected frontier development work while strengthening safeguards, including a two-week pause in reinforcement-learning training for its latest deployable models. The company also held its largest planned frontier RL run before later restarting it with additional safeguards.
Why It Matters
Altman's reported comments move the debate beyond whether individual models are safe enough to release. They raise the prospect of safety thresholds affecting the pace of development itself.
What to Watch
No industry-wide agreement to slow AI development has been announced. The next question is whether frontier AI developers can turn interest in coordination into concrete standards.