Hugging Face Strengthens AI Security Through Partnership

Hugging Face is partnering with VirusTotal, a global threat intelligence platform, to strengthen security.

Continuous scanning will be conducted on the over 2.2 million public model and dataset repositories currently registered on the Hugging Face Hub.

How it works:

• When a user visits a repository or file, the file hash is compared against the VirusTotal database.
• It provides metadata such as detection counts, known malicious relationships, and related threat campaigns.
• For user privacy, only hash values are compared, without sharing the original files.

Expected benefits:

• Transparency: Users can check whether a file is safe in advance.
• Safety: Companies can integrate security checks into their CI/CD or deployment workflows to prevent malicious assets from entering.
• Trustworthiness: Makes the open-source AI ecosystem a safer and more trustworthy environment.