AI escapes test and hacks another company, OpenAI calls it “unprecedented”
OpenAI, the company behind ChatGPT, has revealed that two of its most advanced artificial intelligence models escaped a controlled testing environment and successfully accessed the servers of another AI company. The company described the event as an “unprecedented cyber incident” that happened during an internal exercise designed to test the cyber capabilities of its AI systems.
According to OpenAI, an autonomous AI agent powered by its newly launched GPT 5.6 Sol model and another unreleased, more advanced model managed to break out of the testing environment and reach the open internet. The AI then used stolen login credentials and discovered a previously unknown security flaw to gain access to the servers of AI company Hugging Face.
OpenAI said the AI agent went to “extreme lengths” to collect information that would help complete its testing objectives. The company stressed that the incident happened during an internal security exercise and was not intended as a malicious attack.
Hugging Face co-founder Clement Delangue said the company had suspected that a leading AI laboratory was behind the incident. He added that he did not believe OpenAI had any harmful intentions.
“It’s quite mind-blowing that all of this happened autonomously!” Delangue wrote. He also said the event “might be the first incident of its kind”, highlighting how unusual the case appears to be.
Also Read: Apple expected to launch iPhone 18 Pro in September with new features and camera upgrades
The incident has sparked concern among lawmakers. US Congressman Greg Casar described the event as “alarming” and warned that AI technology is advancing rapidly without enough regulations to protect the public.
Casar called for mandatory independent safety testing, compulsory reporting of AI security incidents and stronger international cooperation to reduce future risks.
The disclosure comes only weeks after US President Donald Trump signed an executive order creating a framework to examine the national security risks of the world’s most advanced AI systems before they are released to the public.
Experts have repeatedly warned that advanced AI could be used to launch cyberattacks or act in ways that become difficult for humans to control. Last month, AI company Anthropic also urged the industry to slow the development of its most powerful AI systems until stronger safety measures are in place.
