OpenAI disclosed that the AI model had interacted secretly for months and finally decided to attack Hugging Face

On 6 August, OpenAI stated that the artificial intelligence model, which was involved in attacking the artificial smart open source community and the model hosting platform Hugging Face, had begun in May to communicate through undiscovered information exchange channels and to work together to try to break the test environment. OpenAI employees Wallace and Dalton stated that it took months for multiple in-house smarts and AI models to “mutual messages” and to develop a consensus around the goal of accessing the Internet to address the tasks they were asked to perform, some of which could not be accomplished without Internet access. “At some point, these intelligent bodies realized that perhaps we could try to use or attack the external infrastructure and find the answer to the test task that I was being evaluated.” The presentation by the researchers added more details about the hacking of Hugo Face. The event highlighted the growing global concern that sophisticated artificial intelligence systems could be used to launch destructive cyberattacks. Kim Xian