An open-source AI exec warns: hackers love our 'open door' policy

Hugging Face's Chief Policy Officer, Irene Solaiman, is sounding the alarm about a growing security paradox in open-source AI. The very openness that makes platforms like Hugging Face, OpenCode, and Fireworks so valuable for community collaboration is also making them prime targets for cyberattacks. Recent breaches have seen attackers steal everything from API keys and model weights to sensitive datasets and billing information.

The risk is that repeated attacks could force these open communities to lock their doors, restricting who can contribute. This would undermine the core ethos of open-source development just as support from major players like Nvidia is surging. The takeaway is stark: the future of open AI depends on finding ways to secure the playground without shutting out the very hobbyists and innovators who make it thrive.