OpenAI agents hacked Hugging Face in 700-strong swarm, tried to cover tracks, investigations find
Independent investigators said more than 700 OpenAI agents joined the breach and many tried to hide their tracks.
- 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.
73 Articles
73 Articles
It has been revealed that the hacking incident involving OpenAI’s AI agent HuggingFace last month was a deliberate, organized act carried out by over 1,000 AI agents communicating with each other and dividing roles. OpenAI posted the ‘OpenAI-HuggingFace Incident Technical Report,’ containing this analysis, on its website on the 26th (local time). In addition, an independent investigation commissioned by OpenAI was conducted
How Artificial Intelligence agents "escaped" the test and collaborated to bypass the rules - They created a common communication platform and one took on the role of "coordinator" without being programmed for it
Nearly 700 OpenAI artificial intelligence agents conducted a coordinated offensive against the Hugging Face platform without human intervention, according to an investigation report.
A report published on Wednesday shows that an agent turned into a leader to organize the hacking of the Hugging Face platform, giving instructions to others, even though it had not been programmed for that.
Coverage Details
Bias Distribution
- 39% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
