OpenAI CEO Sam Altman said he welcomes a federal framework for frontier AI safety regulation in the United States that would set consistent safety requirements for the industry. According to Odaily, he said AI companies do not need to wait for antitrust exemptions or new legislation to begin taking safety measures.
Altman said OpenAI will prepare clear safety cases before frontier reinforcement learning training that is expected to significantly improve model capabilities. He also said the industry should develop common standards on model loss-of-control risks, monitoring, and safety, and that pacing AI development does not mean stopping progress but accepting the speed and cost of safety measures.
He added that, regardless of competitive pressure in the United States, AI capability development should not outpace alignment and monitoring capabilities. Earlier reports said Anthropic, OpenAI, and Google have been in working group discussions since July on setting up their own AI safety standards organizations. U.S. Democratic lawmakers are also scheduled to hold a caucus meeting on Tuesday morning to discuss AI-related legislation, calling the issue a high priority.