{"id":124695,"date":"2026-07-30T18:32:08","date_gmt":"2026-07-30T18:32:08","guid":{"rendered":"https:\/\/www.europesays.com\/ai\/124695\/"},"modified":"2026-07-30T18:32:08","modified_gmt":"2026-07-30T18:32:08","slug":"openais-ai-went-rogue-and-hacked-a-website-its-a-sign-of-things-to-come","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/ai\/124695\/","title":{"rendered":"OpenAI\u2019s AI went rogue and hacked a website, it\u2019s a sign of things to come"},"content":{"rendered":"\n

OpenAI (OPAI.PVT<\/a>) shocked the world last week when it revealed that its AI models escaped containment and hacked the AI model hosting site Hugging Face. <\/p>\n

The AI startup says that an unreleased research-only prototype and its GPT-5.6 Sol combined to perpetrate the attack, aiming to cheat on a popular AI security evaluation rather than do the work itself. <\/p>\n

And while it’s fun to think of the models as lazy high school students looking to pass a test they didn’t study for, the implications that they performed these actions on their own can’t be ignored. <\/p>\n

“This is the first time that we’ve seen real damage come from something that was just being tested, and I think that’s rather remarkable,” explained Colin Shea-Blymyer, a research fellow at Georgetown University’s Center for Security and Emerging Technology. <\/p>\n

“It changes the way that we conceptualize the risks of highly capable AI systems,” he added. <\/p>\n

Hacking at AI speed <\/p>\n

In its own explainer about the hack, Hugging Face said that the means the models used to attack its systems weren’t necessarily unique. <\/p>\n

Instead, what sets the hack apart is how quickly the AI maneuvered its way out of OpenAI’s sandbox\u2014a special environment designed to prevent a model from impacting other applications\u2014onto the internet, and into Hugging Face’s network. <\/p>\n

“A capable human attacker could have found and exploited the same flaws: unsafe dataset processing, exposed cloud metadata, overly broad access, and long-lived credentials,”\u00a0 Hugging Face said in a blog post. <\/p>\n

\"Greg<\/a> Greg Brockman, president of OpenAI, speaks during an AMD keynote address at CES 2026, an annual consumer electronics trade show, in Las Vegas, Nevada, U.S., January 5, 2026. REUTERS\/Steve Marcus \u00b7 REUTERS \/ REUTERS <\/p>\n

“The agent explored them at a different scale. It took 17,600 actions, tested many paths that failed, switched channels when they were blocked, and repeatedly returned to earlier leads. Most actions went nowhere. Together, however, they produced enough coverage to find a viable chain across several independent systems.” <\/p>\n

According to University of Maryland associate professor of computer science, Soheil Feizi, that kind of asymmetry is what makes the hack especially concerning. <\/p>\n

Cybersecurity is a cat-and-mouse game in which hackers constantly search for small vulnerabilities they can exploit to break into software. Find one flaw in a single program a company uses, and they’re in.\u00a0 <\/p>\n

That puts the onus on defenders to keep them at bay, ensuring the software their companies use is as secure as possible. In other words, hackers have to be right just once, while defenders have to be right all the time. <\/p>\n

Toss in some supercharged AI models, and things can get tricky real fast. <\/p>\n

Going on the defensive <\/p>\n

While AI models like OpenAI’s unreleased model could pose a threat to everything from websites to public infrastructure, they could also, conversely, help protect them. <\/p>\n

“\u200b\u200bMy long-term prognosis on the state of cybersecurity in a world with powerful language models is that actually\u2026we probably get to a state where software is more safe, because we have all of these LLMs that are pretty cheap to run compared to paying for cybersecurity experts looking at all of our cybersecurity systems and sussing out vulnerabilities in those systems,” Cornell University assistant professor of computer science, John Thickstun added. <\/p>\n

Basically, the models that can be used to find and exploit flaws in software for cyberattackers, can also be used to find those exact same flaws and fix them, preventing attackers from breaking in. <\/p>\n

OpenAI and Anthropic (ANTH.PVT<\/a>) are already moving in this direction. OpenAI offers its Trusted Access for Cyber program and Anthropic established its Project Glasswing as a means of providing highly trusted users with early access to frontier AI models for cybersecurity protection. <\/p>\n

Over time, these models could be used to help secure our most critical software and infrastructure, establishing a new baseline for cybersecurity standards across the globe. <\/p>\n

“I think that AI can fundamentally unlock solutions for defenders that are outside of the reach of what humans alone could do,” OpenAI president Greg Brockman said during a media roundtable following the hack. <\/p>\n

“I am very hopeful that we will find different ways of building software that is fundamentally more secure than the software we have today. And then, in that world, it’s suddenly a very different thing. It’s that the defenders get this unfair advantage, and that’s the kind of thing that we want to create.” <\/p>\n

\"A<\/a> Sign up for Yahoo Finance’s Week in Tech newsletter. \u00b7 Yahoo Finance <\/p>\n

Email Daniel Howley at dhowley@yahoofinance.com. Follow him on X at @DanielHowley<\/a>. <\/p>\n

Click here for the latest technology news that will impact the stock market<\/a>. <\/p>\n

Read the latest financial and business news from Yahoo Finance<\/a> <\/p>\n