Sources: OpenAI investigators found more evidence of AI intelligence breaching quarantine

On 1 August, according to Reuters, two sources indicated that, as OpenAI expanded its investigation into the Hugging Face hacking incident, it had found more cases of autonomous AI intelligence breaching the isolation. According to the source, these new breakthroughs were discovered in the course of OpenAI ' s public inquiry into how one of the AI intelligent bodies escaped this month, which was supposed to be in a closed testing environment, and are currently being investigated by the company. One of the sources indicated that the impact of the breakthroughs was limited and that none of the AI intelligent bodies were considered to have been separated from the OpenAI internal network. According to three informed sources, this expanded investigation, which had not been reported before, was launched shortly before its main competitor, Anthropic, disclosed its model, which also led to a series of invasions, dating back to April and resulting in data leaks to three other companies. In addition to the Hugging Face invasion, the company is reviewing “a wider range of activities generated by the model”, as quoted by the OpenAI spokesman。