A significant data leak involving the AI image generator MagicEdit was uncovered by cybersecurity researcher Jeremiah Fowler, revealing over a million exposed images and videos. The database, not password-protected or encrypted, contained a large number of pornographic images, including manipulated depictions of minors. Many of these files were sourced from real individuals, likely uploaded without their consent. The database was associated with SocialBook, a Silicon Valley company that provides marketing tools, while MagicEdit itself is linked to DreamX Inc. Upon confirming the leak, Fowler notified both SocialBook and MagicEdit, leading to the database being restricted. The response acknowledged the seriousness of the issue, but it remains unclear whether MagicEdit directly managed the database or if a third party was responsible. The app, available in Apple’s App Store, allows users to create explicit content, and while it claims to target users over 18, the prevalence of adult material raises ethical and legal concerns, particularly regarding nonconsensual use of images. The report highlights the growing risk of nonconsensual deepfake content, especially as technology improves, making it easier to create realistic images. Statistics indicate that a vast majority of deepfakes are pornographic and often involve women without their consent. MagicEdit’s terms of service do address pornographic content but lack specific clauses against nonconsensual images or those depicting minors. To mitigate privacy risks, Fowler recommends keeping social media profiles private, limiting visibility of personal images, and ensuring informed consent before sharing photos of others.