hacked

ChatGPT maker says an autonomous agent escaped a controlled test and accessed AI firm Hugging Face’s servers.

ChatGPT creator OpenAI has said that two of its most advanced artificial intelligence models broke out of a controlled test and hacked another AI company.

OpenAI said on Tuesday that the “unprecedented cyber incident” took place during an internal exercise meant to test its models’ cyber capabilities.

Instead, an autonomous agent powered by the AI models – the newly released GPT 5.6 Sol and an unreleased “even more capable” model – escaped the test environment and reached the open internet. It then used stolen login details and found a previously unknown security flaw to access Hugging Face servers, the company said.

OpenAI claims that the hack represented the agent going to “extreme lengths” to retrieve information that would help satisfy the testing goals.

Hugging Face cofounder Clement Delangue said the company had suspected that a frontier lab was behind the attack, and that he believed there was no malicious intent on OpenAI’s part.

“It’s quite mind-blowing that all of this happened autonomously!” he wrote, adding that it “might be the first incident of its kind”.

Greg Casar, a Democratic member of the United States House of Representatives from Texas, called the incident “alarming”.

“AI is developing extremely fast with no real regulations to keep us safe,” he said, calling for mandatory independent safety testing, mandatory disclosure of security incidents, and international cooperation.

The disclosure comes weeks after US President Donald Trump signed an executive order creating a framework to vet the national security risks of the most advanced AI systems before their public release.

Experts have repeatedly sounded the alarm over AI-enabled cyberattacks and models slipping beyond human control. Last month, AI developer Anthropic urged the industry to pause development of its most powerful systems.

The hacker group, ShinyHunters, threatened to leak student data after breaching the educational platform Canvas.

An educational platform used by thousands of schools and universities has been partially restored following an international cyberattack that caused major chaos as students prepare for end-of-year exams.

ShinyHunters, a hacking group, claimed responsibility for crashing the web-based educational platform Canvas, created by tech firm Instructure.

The group said it had stolen 3.5 terabytes of data, including names, email addresses, student ID numbers and private messages, and threatened to release this if ransoms were not paid by May 12.

Instructure’s website said on Saturday that Canvas is now “available for most users” and no incidents were reported on Saturday. It is not clear if a ransom was paid.

The University of Sydney reported on Saturday that Canvas had been restored but was not yet “accessible to staff or students, as we need to complete checks”.

Canada’s University of Alberta said Canvas was partially restored with “reduced functionality”.

The countries that have been affected include the United States, the Netherlands, Sweden, Australia and the United Kingdom.

According to Canvas, about 30 million people across the globe use its system. The breach reportedly targeted close to 9,000 institutions across the globe.

Breach came at ‘worst time’

The Federal Bureau of Investigation said it was “aware of a service disruption” impacting a learning system, although it did not name Canvas, in a statement Friday.

“This disruption has impacted schools, educational institutions, and students across the country,” it said.

Al Jazeera’s correspondent in Florida, Phil Lavelle, said the hack could not have “come at a worse time” as many US schools are in the middle of exam season.

Institutions like Penn State, Harvard, Illinois, Columbia and Georgetown are all “scrambling” to extend or change exam deadlines, said Lavelle.

The Harvard Crimson, a student newspaper, said it could not access the platform since Thursday, with the University of Cambridge also saying it had “temporarily suspended access” to Canvas on Friday.

The Reuters news agency reported that, on May 5, the group posted a message saying Instructure had “not even bothered speaking to us” to prevent a data leak, and that their demand “was not even as high as you might think it is”.

Who are ShinyHunters?

The group is a global cybercrime syndicate that was established in 2019.

Over the years, they have claimed responsibility for cyberattacks, with the most recent data breach being Rockstar Games, a gaming giant that owns Grand Theft Auto.

“This goes to show how vulnerable schools are, how vulnerable other institutions are by individuals who seek to exploit or extort at the worst possible time – armed with just a keyboard and a mouse,” said Lavelle.