Open-source AI platforms are under threat from cyberattacks, according to Irene Solaiman, chief policy officer and SVP of strategy at Hugging Face.
She points to recent incidents like OpenCode's financial losses after hackers exploited API keys in October 2026, and a separate attack where an OpenAI agent swarm accessed sensitive Hugging Face data.
The big worry: these attacks could make open-source AI less accessible for everyone.
Hugging Face, GitHub tighten security
To fight back, Hugging Face is stepping up its security, rotating admin access more often and tightening network controls, while still trying to keep things open for researchers.
GitHub is also making changes by revoking old tokens and making it tougher to publish packages.
As Solaiman puts it, the real challenge now is keeping AI both secure and open so independent developers can still contribute without putting the community at risk.