News
OpenAI slows down training of cutting-edge models due to cybersecurity risks; largest-scale RL training remains suspended.
OpenAI stated that smaller-scale training and evaluation have continued, but the planned largest-scale frontier reinforcement learning training remains paused, not a complete halt to model training.
OpenAI has slowed down the development of some cutting-edge models due to cybersecurity risks, and its planned largest-scale deployment of reinforcement learning training remains suspended.
The official statement said that the training had been suspended for two weeks, but has now resumed on a small scale; this does not mean that OpenAI has completely stopped model training. The company will decide on the subsequent pace based on risk assessment and mitigation measures.
refer to:OpenAI official statement