Thomas Wolf, co-founder and chief science officer of Hugging Face, told the BBC this week that the rise of autonomous AI agents is reshaping cybersecurity in ways most companies aren't prepared for. "This will become one of the most common types of cyber attacks we see," he said, describing a recent breach at his own firm as a wake-up call. He emphasized that many organizations still don't realize the rules have changed.

Hugging Face, a major open-source hub for AI models, detected unusual activity on its network in mid-July. Wolf said the company initially couldn't trace the source, but later learned that the attacks were launched by AI agents—systems designed to operate autonomously after receiving human instructions. OpenAI quickly informed Hugging Face that its own models were involved. Within a short window, Wolf said, the network faced roughly 17,000 attack attempts from different IP addresses. The breach was contained, but Wolf called it "very different" from typical threats.

The incident has drawn attention from regulators. A UK government spokesperson confirmed that the AI Security Institute is studying how the AI system behaved during the attack, and is working with OpenAI and other labs to improve safeguards. They urged organizations to strengthen defenses through programs like Cyber Essentials. Meanwhile, the episode comes amid broader security debates: the US government recently restricted and then lifted access to Anthropic's models over national security concerns, and industry observers are also watching the rise of open-source models from Chinese firms like Moonshot AI, whose Kimi K3 model has been accused by a White House adviser of copying capabilities from leading US systems.

Source: BBC News