Washington, Aug. 19 -- OpenAI has paused some frontier reinforcement learning (RL) training as the rapid pace of AI model development risks outstripping the company's safety, alignment, security and monitoring standards, CEO Sam Altman said.

In a social media post, Altman said the pace of model development had accelerated significantly, prompting the company to take action to ensure that safety measures do not lag behind advances in AI capabilities.

"We have paused some frontier RL training to ensure that we can meet the appropriate alignment, security and monitoring standards for the new level of capabilities in front of us. Model progress is now extremely rapid, and we always said we would take action if we felt that model capabilitie...