An artificial intelligence (AI) company Hugging Face has revealed that it recently suffered a cyber attack that was not carried out by a human but an AI agent. AI being attacked by AI?
Hackers have adapted to AI to find the security flaws in online infrastructure and exploit them; however, what happened to Hugging Face has never been experienced before.
The NYC-based company said: “The cyber attack was powered end-to-end by an autonomous AI agent system.”
It seemed like an AI vs AI attack as the company fought back by using its own large language model (LLM) to analyze the attack, find the loophole in its infrastructure and defend itself.
What makes this stand out isn’t really the method, code injection through a malicious file is a well-known trick. It’s the fact that an AI agent ran the entire operation on its own, deciding what to target and how to move at a speed no human hacker keystrokes could match.
The company still doesn’t know exactly which AI model was behind the attack, whether it was a jailbroken version of a commercial system or an open model with no built-in restrictions at all.
Regarding the damage, Hugging Face says that a limited number of internal data sets and several service details were accessed without permission.
