Malicious Hugging Face Repository Typosquats OpenAI (infosecurity-magazine.com)

• HiddenLayer discovered infostealer in a Hugging Face repository typosquatting OpenAI.
• The repository stole credentials and API keys from developers.
• The incident highlights security risks in open-source AI repositories.

"HiddenLayer has uncovered an infostealer malware in a Hugging Face repository that typosquats OpenAI. The malicious code steals credentials and API keys from developers who download the model. This discovery highlights cybersecurity risks in AI model repositories and the need for vigilance in open-source AI supply chains."