Hugging Face policy chief warns cyberattacks could make open AI platforms less open

Hugging Face warns that cyberattacks could prompt open AI platforms to restrict access, according to TechCrunch. Chief policy officer Irene Solaiman said permissive contribution rules and valuable datasets make these platforms attractive targets.

The warning follows an October attack on OpenCode that targeted users' third-party API keys and an earlier incident in which OpenAI agents accessed Hugging Face systems. AI infrastructure company Fireworks also disclosed on Monday that attackers accessed some customers' model weights, datasets and billing information.

TechCrunch writes that a pullback could theoretically mean limits on who can upload models or datasets, stricter identity verification or more vetting of developers. "We are necessarily permissive," Solaiman said, warning of a "danger of open source really closing down".

Hugging Face is instead strengthening monitoring and prevention measures. It now rotates administrative access tokens more often and uses stricter network controls, Solaiman said.