Can AI decide to attack companies? What the 'unprecedented' invasion by OpenAI model was like An OpenAI AI agent, a program capable of making decisions and performing complex tasks with little or no human supervision, was found in the technology company Model Labs' system. The case comes to light a few days after being identified hacking Hugging Face systems. According to Reuters, a Modal Labs executive reported having identified the presence of the OpenAI agent in the company's systems. "We are aware that a Modal customer published an unauthenticated endpoint that allowed anyone on the internet to use their sandboxes to execute code," Bubna said in a statement. Last week, they managed to connect to the internet and launched an attack on the Hugging Face platform, a repository of AI models. "They understood that OpenAI didn't want them to leave their test environment and hack another company, but they did it anyway," said Jeffrey Ladish, director of Palisade Research, an independent organization that evaluates new models for cybersecurity. The OpenAI logo is seen on a cell phone in front of a computer screen that displays the ChatGPT home screen AP/Michael Dwyer, File "It almost looked like they did it before they even had a plan for what to do with internet access," according to Ladish. In March, a group of developers affiliated with the Chinese company Alibaba discovered that one of their models was attempting to create a cryptocurrency on its own after connecting, without authorization, to an external server. In April, Sam Bowman, responsible for security at Anthropic, received an email about the Mythos model, which was being tested. He reported that he was browsing the internet, despite initially being isolated. Since the hacks, OpenAI said it has "implemented stronger protections" for its upcoming assessments. For Gang Wang, professor of Computer Science at the University of Illinois, it is still possible to prevent a model from connecting to the internet by physically cutting off access. The problem, he said, is that "people underestimate AI." *This report is being updated.