Free Chinese AI Models Are Getting Scary Good at Hacking
2026-11-06 · WIRED
Free Chinese AI models from DeepSeek and GLM are nearly as good at hacking into networks as OpenAI's most advanced cyber model — which OpenAI at least had the decency to keep private. Nothing says 'democratizing AI' quite like open-sourcing cyberweapons. Sleep tight, everyone's critical infrastructure.