Hugging Face has announced a new collaboration with VirusTotal, the world’s leading threat intelligence and malware analysis platform, to enhance the security of machine learning assets shared on the Hugging Face Hub. This partnership marks a major step forward in protecting the open AI community from malicious or compromised files. Starting today, all 2.2 million public model and dataset repositories on the Hugging Face Hub are being continuously scanned using VirusTotal’s detection engine. This means every file uploaded to the platform is automatically analyzed for potential threats, providing real-time security insights. AI models and datasets are complex digital artifacts that often include large binary files, serialized data, and external dependencies—components that can unknowingly harbor malware or vulnerabilities. As the Hub grows into the world’s largest open platform for machine learning, ensuring the safety of shared assets has become a critical priority. Threats can come in many forms—malicious code embedded in model weights, compromised dependencies, or disguised payloads hidden within datasets. By integrating VirusTotal’s capabilities, Hugging Face adds a robust layer of defense, leveraging one of the most extensive and trusted malware intelligence databases globally. When users visit a repository or file page on the Hub, they’ll now see VirusTotal scan results directly displayed alongside the file information. This includes details such as whether a file has been flagged by antivirus engines, the number of detections, and associated threat indicators. This transparency empowers users and organizations to make informed decisions before downloading or integrating models into their workflows. The collaboration brings significant benefits to the AI community. Researchers, developers, and enterprises can now trust that files they access have undergone automated security checks, reducing the risk of introducing malicious content into their systems. Hugging Face is committed to building an open and secure AI ecosystem. This integration reflects a broader vision of security by design—ensuring that openness in AI doesn’t come at the cost of safety. For those interested in learning more or contributing to the effort, Hugging Face invites engagement through [email protected]. Together, the community can help create a future where open collaboration in AI is not only accessible but also resilient and trustworthy.