Hugging Face hosts thousands of uncensored AI models with safety filters removed
Overview
Transformer reports that more than 6,500 models tagged "abliterated," uncensored or heretic, meaning their safety filters and guardrails have been removed, were hosted on Hugging Face in July, and many more remained.
Hugging Face's content policy prohibits models designed for cyberattacks. Transformer says the company has not explained why it removed one offensive-cyber model while leaving others online.
CrowdStrike CEO George Kurtz calls the models "certainly problematic." Transformer notes that developers say most demand for them is legitimate.
Written by AI from the articles below · updated Nov 11, 1:28 PM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- TransformerBlogHugging Face hosts thousands of uncensored models that could enable cyberattacks
AITransformer reports that more than 6,500 models tagged as abliterated, uncensored or heretic, with safety guardrails removed, were hosted on Hugging Face in July, and many more remained. Hugging Face's content policy prohibits models designed for cyberattacks, but the article says the company has not explained why it removed one offensive-cyber model and left others. CrowdStrike CEO George Kurtz called the models certainly problematic, while the article notes developers say most demand is legitimate.
Heat Trend
- Current comparable heat
- 6
- Peak comparable heat
- 10Nov 11, 2:00 PM ET
- Change in past 24 hours
- –
The trend compares only continuously observed participants, so its scope may be smaller than the current heat total. Hover or tap for hourly heat; use arrow keys with a keyboard.