**Source : TIME | 4 août 2026** "AI models are unpredictable and their risks scale with their capabilities," writes Garrison Lovely. 👉 **Lire l'article complet sur TIME :** [https://time.com/article/2026/08/04/what-openai-s-hugging-face-hack-tells-us-about-ai-s-risks/](https://time.com/article/2026/08/04/what-openai-s-hugging-face-hack-tells-us-about-ai-s-risks/)