The lineage behind 69% of open models was never verified. Cisco just fingerprinted almost 900 for free
AI Summary
Cisco's security team has fingerprinted nearly 900 open-source models to verify their lineage, highlighting a gap in the verification process currently utilized by platforms like Hugging Face. This initiative aims to enhance the security and trustworthiness of open models, which often lack verified documentation regarding their origins.
From the source
A security team approving an open-source model for production today starts with a repository page. The page lists the model name, the license, and a tag identifying the base model it descended from. That tag is a string the uploader typed. Hugging Face does not require uploaders to substantiate the claim through weight-level analysis. The ATOM Report, published by Nathan Lambert and Florian Brand at Interconnects AI in April 2026, tracked roughly 1,500 mainline open models. ATOM identifies deriv
The full text couldn't be loaded here (the source may require a subscription).
View original at VentureBeat AI