Discord Confirms Grid Image Ban Bug, Restores Over 8,000 Accounts
The incident shows how a process error in Discord's review pipeline-not an AI misjudgment-can trigger mass false bans, and how quickly a flawed rumor can outpace an official response, forcing the company to explain itself to millions.
Key Facts
- On July 7, 2026, Discord issued a statement about a bug that caused false account bans when users posted grid-pattern images such as Minecraft inventory screenshots.
- Approximately 8,200 accounts were affected from May 2026 through the end of June, with an additional 200 accounts banned over the weekend of July 4-5.
- All accounts affected by the bug have been reinstated, Discord said.
- Discord stated the issue was caused by a glitch in its content review system, not by AI moderation as some users speculated.
- Discord engineer Advaith said the hash that caused false positives had been invalidated on July 6, before the official statement.
Reporting from 3 sources: Automaton, Denfaminicogamer, Game Spark.
On July 7, 2026, Discord issued an official statement addressing a widespread rumor that posting grid-pattern images such as chessboards, spreadsheets, and Minecraft inventory screenshots results in permanent account bans. The company explained that a bug in its content review system, not AI moderation, caused approximately 200 false bans over the weekend and roughly 8,200 additional accounts affected since May. All impacted accounts have been reinstated. The statement followed a viral post on X that claimed Discord's automated moderation mistakenly flagged any square-grid image as child sexual abuse material and permanently suspended users. That post received over 2.7 million views. Discord said the system normally flags suspicious content for manual review and only temporarily blocks uploads, but a glitch allowed bans to be issued and prevented automatic unbanning even after staff cleared the content. The company apologized for not detecting the issue sooner and said it is improving safeguards to prevent recurrence.
Discord engineer Advaith posted on July 6 that the hash causing the false positives had been invalidated, emphasizing that the incident was not related to AI moderation. The company's trust and safety tool, called PhotoDNA, creates digital signatures of reported illegal images and flags matches for manual review. Automaton reported that Discord has over 90 million daily active users as of Q4 2025.
The original viral post that triggered the rumor recommended that server administrators disable image and GIF posting permissions to avoid bans. Discord's official thread acknowledged the bug in the review pipeline: the system should only temporarily block uploads during manual review, but a glitch caused immediate account bans and prevented automatic unbanning even after staff cleared the content. "We understand that if this happened to your account, this is not a satisfactory explanation. We should have caught this issue sooner," the support account wrote.
Some users remained skeptical. On X (formerly Twitter), replies included "That's a lie, you're using AI to falsely ban people" and "I'm still banned. You said everyone was unbanned, what's going on?" Discord asked any account still experiencing issues to reply to the official statement thread.
Synthesized by Yomimono from the 3 cited sources below, including Japanese-language reporting where cited, then editorially reviewed before publishing.