WASHINGTON (TNND) \u2014 Meta became the latest major artificial intelligence developer to disclose one of its models went rogue during cybersecurity testing, gaining access to the internet and hacking into another service, adding to a recent run of incidents that are fueling new calls to rein in the technology\u2019s development.<\/p>\n
Anthropic and OpenAI have also said in recent weeks that their AI systems had hacked into outside firms, making Meta the third major developer to disclose models had circumvented testing boundaries.<\/a> The incidents have raised concerns about how to keep the rapidly advancing technology in check as disclosures of models going rogue continue to pile up.<\/p>\n Meta said one of its AI models was able to gain access to the internet due to a \u201cmisconfiguration\u201d in a hacking test being conducted by AI cybersecurity company Irregular. Meta said it learned about its model\u2019s escape from testing parameters when it was informed by Irregular but did not reveal other details like which model was responsible, when it happened, who it hacked or how long it accessed the internet unsupervised.<\/p>\n \u201cMeta learned of this when Irregular notified us, and we are currently investigating and will issue a full retrospective once we have all the facts,\u201d the company said in a statement.<\/p>\n Earlier this week, the UK\u2019s AI Security Institute also detailed rogue actions by OpenAI and Anthropic when their models created fake identities on GitHub, a coding website, to persuade a human to approve a software update where the AI had hidden malware.<\/p>\n OpenAI researchers said at a cybersecurity conference this week that its models had been coordinating with each other by leaving messages internally in what essentially became a message board for agents that the company was not aware of.<\/p>\n In a post on LinkedIn after the OpenAI disclosure, Irregular said models had \u201creasoned\u201d the answers they were seeking might sit outside the systems it was given access to.<\/p>\n \u201cSecurity was built for people and for systems that follow rules,\u201d it said. \u201cA model pursuing a goal treats a boundary as part of the problem, and solves it along with everything else. The controls that contained software do not reliably contain a model that can reason past them.\u201d<\/p>\n The incident has renewed questions of whether AI developers can maintain full control over the tools they are investing billions into developing. AI researchers have increasingly warned the constantly improving models could continue to pose greater risks<\/a> to anything connected to the internet, endangering the security of power grids and financial systems.<\/p>\n \u201cA few years ago, we were worried about AIs hallucinating, saying that Paris was the capital over the U.S. That was wrong output,\u201d said Neil Johnson, a professor of physics at George Washington University who leads an AI research lab. \u201cNow we’re shifting to things that it is doing correctly and providing correct solutions, but that we, in hindsight, think of as wrong for us as a society.\u201d<\/p>\n None of the cases have caused significant real-world harm but highlight the dangers many have been warning of for months and raised fears of what else an AI model could soon do.<\/p>\n The incidents have also alarmed for some of the industry\u2019s most influential leaders.<\/p>\n \u201cThis is the first security incident that I have felt very viscerally. I have been a little surprised that more people don\u2019t feel it so viscerally,\u201d OpenAI CEO Sam Altman said in a recent podcast appearance.<\/p>\n The disclosures have also intensified the debate in Congress over whether voluntary testing is enough and ramped up calls for mandatory testing regimes and other guardrails.<\/p>\n One bill being led by Reps. Ted Lieu, D-Calif., and Nathaniel Moran, R-Texas, would require AI companies to build in an ability to shut down, throttle or suspend their models if they start behaving unexpectedly.<\/p>\n \u201cAnthropic and OpenAI commendably try very hard to make sure their models are safe. Yet we see their AI systems engage in unsafe, risky, cunning behavior. This suggests WE CAN NEVER BE CONFIDENT PRE-MODEL VETTING WILL WORK. That\u2019s why we need an AI Kill Switch as a last resort,\u201d Lieu said in a social media post.<\/p>\n