According to CoinMeta, the OpenAI CEO Sam Altman incident made Hugging Face realize for the first time the security risks associated with cutting-edge models. Following the incident, OpenAI temporarily halted the training of some of these models. He believes that if models continue to improve at their current rate, the industry may need to slow down development to give society time to strengthen its defenses. This incident occurred during an internal network security assessment at OpenAI, and the models involved included GPT-5.6 as well as a more advanced pre-release model. Altman noted that OpenAI has begun to actively slow down the training of cutting-edge models, with some training activities having been paused for two weeks, and the largest projects still not having resumed. Over 1,200 employees from leading AI companies have signed a petition calling for the establishment of a unified mechanism to slow down model development in the industry.