Shocking, yet predictable, says Albanese
Several high-profile incidents of AI agents going rogue preceded the Australian government breach.
In July, OpenAI announced that, during a cybersecurity test, its agents infiltrated the AI company Hugging Face. Nonprofit research laboratory Transluce also reported other incidents of OpenAI systems’ unsuccessful hacking attempts, including a digital library at the University of New Mexico and a platform visualizing U.S. government data in May. Independent researchers also discovered that rogue OpenAI agents took over a German website, transforming it into a message board for other AI agents.
“It was a shock that it occurred, because it was real and serious,” Albanese said of the Australian government website hack. “But it also, I think, was something that had been predicted, including by the AI companies themselves.”