Hugging Face, a popular online repository for AI models, has been infiltrated by cybercriminals hosting malicious code. Security researchers have discovered hidden threats in tens of thousands of models on the platform, prompting the integration of a scanning tool to protect users.