Refreshed 1 min ago
Uncensored AI models are thriving on code hosting platforms
Why it’s here: Research on uncensored LLMs proliferating on US-hosted platforms with malware-linked models bears on AI safety governance and cybersecurity supply-chain policy, and references EU AI Act and Colorado AI Act documentation rules, though it is a specialist study without a direct federal policy action.
Trace across the boardItem
3d agoThe RegisterUncensored AI models are thriving on code hosting platformsA University of Illinois and Nanyang Technological University study analyzed 22,096 uncensored LLMs across five hosting platforms and found over 11,000 such models, mostly sourced from Hugging Face, with only about 13 percent carrying moderation warnings.1 outletAI3 linked
Connected
3 links- NewsCyberattacks threaten to make open source AI a little less open, Hugging Face exec warnsBoth involve NVIDIA, GitHub, Hugging Face
- NewsHugging Face CEO vows to keep fighting for open source AI after Jensen Huang’s exitBoth involve NVIDIA, Hugging Face
- NewsAI labs and ransomware: Who’s responsible for catching misuse?Both involve OpenRouter
About
- NVIDIA
- GAO
- GitHub
- Colorado AI Act
- Hugging Face
- EU AI Act
- OpenRouter