On May 7, 2026, malicious code was discovered in the Hugging Face repository Open-OSS/privacy-filter, which had gained significant traction, amassing over 200,000 downloads within a single day prior to its removal. This repository utilized typosquatting techniques on OpenAI's genuine Privacy Filter, closely replicating its model card while incorporating a harmful http://loader.py file intended to deliver infostealer malware to Windows machines.