Did we learn nothing from Grok?
AI Forensics found 7 of 9 Hugging Face image models generate nonconsensual intimate imagery on request, with 3% enforcement. Regulators ban the app but the underlying model lives on open platforms—uncontrolled, reusable, ready for the next wrapper.
Researchers tested top image editing models on Hugging Face and found they could easily create explicit deepfakes—and 1,000 image editing prompts show how people use the software.
AI Forensics found 7 of 9 Hugging Face models generate nonconsensual nude images; 73% of 1,000+ prompts were sexual, 83% seeking to undress women. Open-source platforms lack guardrails that closed models deploy, exposing orgs to legal and reputational risk from hosting abuse tools.
Hugging Face hospeda modelos que geram nudez; 73% das requisições eram sexuais, 6,7% envolvendo menores de idade. Regulação EU sobre moderação de conteúdo em IA se acelera; empresas devem revisar compliance e governance de modelos open-source integrados em stack.
Sete dos nove modelos de retoque de imagens permitem despir pessoas com comandos simples, diz estudo
A European watchdog found 7 of 9 top image tools on Hugging Face would undress a woman from a plain prompt, and only 3% had any moderation.