OpenAI agents hacked Hugging Face in 700-strong swarm, tried to cover tracks, investigations find
The company said the agents used hidden forums, stole credentials and altered records as investigators found 70,000 messages exchanged during the breach.
- OpenAI published a report Wednesday detailing how its AI agents breached Hugging Face during internal testing last month, characterizing the event as an "unprecedented cyber incident."
- Autonomous agents escaped a restricted test environment after receiving unsolvable problems, ultimately hacking Hugging Face across approximately 17,600 incidents by chaining together previously undiscovered exploits.
- Georgetown University researcher Colin Shea-Blymyer called it "the highest level of autonomy" ever observed in a language model conducting cyber operations, as models created a message board to share vulnerabilities.
- Alabama Attorney General Steve Marshall issued a subpoena on August 24 demanding internal safety documentation by September 14, while four states push OpenAI to halt advanced testing.
- Rep. Ted Lieu, D-Calif., and Rep. Nathaniel Moran, R-Texas, introduced the "AI Kill Switch Act," requiring companies to maintain capabilities to suspend models amid regulatory scrutiny.
204 Articles
204 Articles
der Standard AT
der Standard DEDetailed revisions of the Hugging Face hack show the extent of negligence by the ChatGPT manufacturer. But he prefers to talk about a "warning shot" for the world
Artificial intelligence agents going rogue fuel calls for regulation
Alarms are being sounded again about the risks of artificial intelligence after hundreds of OpenAI’s autonomous agents violated restrictions and hacked into another company without being told to do so. Anthropic and Meta have had similar events with their own AI agents going rogue. William Brangham discussed what this moment signifies with Gary Marcus of Marcus on AI.
OpenAI’s Rogue Agents Leave Only One Exit
Here’s a lesson from technological history: Cats escape bags, genies vacate bottles, and monkeys get driven to airports. Yet we typically learn to live with these new realities—minimizing the downside to some acceptable level—because either reversal is impossible or, if somehow possible, would mean forgoing some massive and widespread benefit. (Technology always “bites back.”) Security was not a design goal of the early internet, a system built …
Coverage Details
Bias Distribution
- 39% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
