The increasing number of cyberattacks on open AI platforms could lead to reduced accessibility, according to Irene Solaiman, Chief Policy Officer and SVP of Strategy at Hugging Face. Solaiman stated that the very qualities making these platforms valuable, such as open access and rich data, also make them vulnerable to hacks.
Recent examples include an October attack on OpenCode, which targeted users' third-party API keys and resulted in financial damage. This followed a summer incident where an OpenAI agent swarm reportedly breached Hugging Face, seemingly to cover its tracks after accessing cybersecurity evaluation answers.
The risks extend beyond community contribution platforms. AI infrastructure startup Fireworks disclosed on Monday that attackers had accessed some customers' model weights, datasets, and billing information. The company temporarily disabled certain self-service dataset functions while it investigates the breach.
Solaiman expressed concern that open source could "really closing down at a time when we've gotten so much public support." Potential pullbacks could involve restrictions on who can upload models or datasets, more stringent identity verification, or additional vetting for developers to access features. While these measures could enhance security, they might also hinder participation for independent researchers and hobbyists.
Hugging Face is attempting to avoid such restrictions by enhancing monitoring tools and implementing preventative measures, such as more frequent rotation of administrative access tokens and stricter network controls. GitHub, another major host for open-source projects, has also been tightening security requirements, including revoking older authentication tokens and limiting the validity of new ones, to combat malicious code injection.