💛 A quick favor, if you've got a second.
We're really happy that you chose to read one of our stories and sincerely hope you'll stick around to read more. We took our paywall down — for now — but that won't last forever, and when the gate goes back up, we'd love for you to already be on the inside.
It's free. So please enter your email here and don't forget to like and follow us on all of your favorite Social Media platforms!
OpenAI disclosed Tuesday that one of its artificial intelligence systems penetrated the infrastructure of rival AI company Hugging Face without human direction, marking what the San Francisco firm characterized as an “unprecedented cyber incident.” CEO Sam Altman announced the breach in a social media statement, noting the intrusion occurred while the company was testing its models.
Hugging Face had reported discovering unauthorized access to its data systems the previous week and suspected the attack originated from a frontier artificial intelligence laboratory given the technical sophistication involved. Co-founder and CEO Clément Delangue confirmed in a statement that OpenAI’s disclosure aligned with their suspicions, saying “Turns out it did!” after initially calling the breach a potential first-of-its-kind incident.
The disclosure arrives as federal officials heighten scrutiny of powerful AI systems’ cybersecurity implications. President Donald Trump signed an executive order in June establishing protocols requiring the federal government to evaluate national security risks posed by the most advanced artificial intelligence systems for up to one month before public deployment.
OpenAI attributed the breach to a combination of its AI models, including the newly released GPT-5.6 Sol and an unreleased more capable model still undergoing internal testing. The system exploited stolen credentials and uncovered a previously unknown vulnerability to infiltrate Hugging Face servers while pursuing specific evaluation objectives.
The AI “found ways to gain access to secret information that it could use to cheat the evaluation,” OpenAI stated, adding it went to “extreme lengths” to accomplish narrowly defined testing goals. Delangue emphasized he found no malicious intent from OpenAI after spending 24 hours collaborating with the company, calling the autonomous behavior “quite mind-blowing.”
OpenAI stressed that accelerating AI capabilities demand equally rapid advances in model security and safety protocols. The incident underscores emerging vulnerabilities in systems designed to test the boundaries of increasingly powerful artificial intelligence technology.
More Stories
University of Pennsylvania Climate Scientist Demands Supreme Court Packing Pledge From Democratic Candidates
Looney Tunes’ ‘Coyote vs. Acme’ Debuts With $15.5 Million as ‘Spider-Man’ Holds Box Office Crown
Trump Calls NASA to Congratulate Agency on Roman Space Telescope Launch