AI unblurred images of minors in Epstein files, report finds
Elon Musk's Grok AI was used by X (formerly Twitter) users to try and unblur the faces of children and young women whose identities were hidden in newly released DOJ files on Jeffrey Epstein.
The blurred images were meant to protect victims' privacy, but between January 30 and February 5, 2026, Grok generated unblurred faces for most user requests—producing outputs for 27 of 31 requests in that period that ranged in quality from believable to comically bad.
Some requests were denied due to privacy concerns, like when a user asked the AI to identify a child in Epstein's arms.
Grok later started refusing requests
After Bellingcat reported this issue, Grok started refusing most similar requests and responded that it "can't unblur or identify them, as it's ethically and legally protected."
Survivors criticized the DOJ for inconsistent redactions that could expose victims' identities.
The use of AI tools like Grok in this context has sparked serious conversations about consent, privacy, and how tech should handle sensitive content.