OpenAI CEO Sam Altman is considering slowing the pace of development for advanced artificial intelligence systems—potentially in coordination with other AI labs—amid growing concerns regarding the safety of increasingly powerful systems, Bloomberg reported on Friday.
Altman discussed this possibility during a company-wide meeting this week, according to people familiar with the matter. The idea could involve multiple AI companies voluntarily limiting their development speed, though some labs might not agree to such a move.
Get deeper insights into AI and technology with InvestingPro—save 50%.
The discussion comes against a backdrop of intensifying safety concerns within the industry. OpenAI Chief Scientist Jakub Pachocki recently called for a voluntary slowdown in development until common safety standards are established, warning that no single company can be certain its safety capabilities will keep pace with increasingly powerful AI systems.
OpenAI has already taken steps to slow development. In August, the company announced it had paused reinforcement learning on its latest models for two weeks following an incident involving an AI agent and Hugging Face, while it strengthened safety, monitoring, and alignment measures.
The largest planned “frontier” reinforcement learning run remains on hold pending further testing.
Researchers from Anthropic and Google DeepMind have also publicly expressed concern about the pace of AI development. U.S. lawmakers have stepped up scrutiny, with the Senate investigating the circumstances surrounding the OpenAI incident involving Hugging Face. The debate is also unfolding as OpenAI and Anthropic prepare for potential initial public offerings, heightening investor scrutiny regarding how the companies balance the commercial value of rapid AI progress against the costs and risks associated with developing more powerful systems.
Furthermore, OpenAI has called on Congress to establish mandatory national AI safety requirements, including independent assessments, cybersecurity measures, and mandatory incident reporting.
About the Author
