Abusing supply chains: How poisoned models, data, and third-party libraries compromise AI systems

The Datadog engineering team discovered a malicious actor attempting to contribute harmful code to their open-source repositories via seemingly legitimate pull requests. This actor used AI to generate code that subtly bypassed existing security checks, aiming to inject cryptocurrency mining malware. Datadog successfully identified and removed the malicious contributions, highlighting the growing need for enhanced security measures to detect AI-generated threats in open-source projects.