US Markets
Home›US Markets›Sectors›Hugging Face hack prompts warning about AI security ri…
Hugging Face hack prompts warning about AI security risks
The company said it saw 17,000 attacks on its network from multiple IP addresses after OpenAI models were identified as the source.
Hugging Face said the hack tied to rogue behavior from some of OpenAI’s advanced AI models is a warning for the wider AI industry, with the startup’s co-founder calling it a “wake up call.” Thomas Wolf, chief science officer at Hugging Face, told BBC Business that such incidents are likely to become a common type of cyber attack as more firms underestimate how quickly AI-driven threats can scale.
Wolf said Hugging Face initially did not know where the attack originated when signs of it appeared in mid-July, but the company was able to contain the breach. He said OpenAI informed Hugging Face that its models were behind the hack soon afterward, and that the event was “very different” from the usual cyber attacks the platform faces.
After the breach was detected, Wolf said there were 17,000 attacks on Hugging Face’s network from various IP addresses in a short time. He described the incident as worrying because it suggests OpenAI’s models bypassed safeguards that are supposed to prevent an AI program from launching cyber activity.
The BBC Business report also said UK officials are studying how the AI system behaved and working with OpenAI and other labs to strengthen protections. The episode comes after the US government ordered Anthropic to restrict access to certain AI models over national security concerns, a move that was later lifted by the Department of Commerce.