A security flaw left an AI image‑generation startup’s cloud storage publicly accessible, revealing over one million images and videos, most of them explicit and many featuring non‑consensual or underage subjects. The breach involved the companies MagicEdit, DreamPal and their parent entity DreamX, as well as related brands BoostInsider and SocialBook. Security researcher Jeremiah Fowler reported the exposure, prompting the firm to shut down access, suspend its products and launch an internal investigation. The incident highlights ongoing challenges around AI‑generated sexual content, child protection and trust‑and‑safety practices in emerging tech companies. Leia mais →