OpenAI found more cases of agent escape, but the complete internal investigation details have not yet been publicly released.

#AIOpenAI More signs of agent escape were found during the internal investigation, but the complete internal investigation details have not yet been publicly released. Reuters obtained information from people familiar with the matter that OpenAI discovered signs of other autonomous AI agents breaking through the isolation environment while investigating the HF face-hugging attack. However, people familiar with the matter said that the newly discovered situation was described as being limited in scope. The agents are currently not believed to have left OpenAI's own network, but OpenAI has included these clues for further investigation. View the full text: https://ourl.co/114151