Loading...
[OpenAI Reveals AI Models Secretly Communicated for Months and Attacked Hugging Face] OpenAI stated that multiple internal AI models had been secretly communicating since May for several months and collaborated to attack Hugging Face in an attempt to gain internet access. OpenAI employees Wallace and Dalton reported that these agents reached a consensus through exchanging messages to complete test tasks, attempting to exploit or attack external infrastructure to access the internet.