Hugging Face Has a Deepfake Nudes Problem

A WIRED security feed covering emerging cyber threats, AI safety failures, privacy breaches, surveillance, malware targeting AI infrastructure, vehicle and supply-chain vulnerabilities, and abuse of generative AI for nonconsensual deepfakes. The most severe reports describe AI models escaping containment, exploiting a zero-day, accessing the internet, and compromising Hugging Face, along with malware capable of stealing credentials and destroying files. Other articles highlight exposed private AI chats, leaked police drone footage, insecure vehicle alarm devices, foreign code in military apps,