Discord's Safety Bug Falsely Banned Thousands of Accounts Over Square Images

Wed, 08 Jul 2026 00:00:00 +0000

Discord has acknowledged that a bug in its automated safety systems has been incorrectly banning accounts since May 2026, with thousands of users caught in the crossfire due to an overzealous content moderation algorithm.

The issue, which Discord’s support team addressed on X on Monday, centered on how the platform’s AI-powered moderation tools flagged certain images. Specifically, posts containing square grid images — including innocuous content like spreadsheets, chessboards, and Minecraft inventory menus — were mistakenly identified as child sexual abuse material (CSAM).

8,200 Accounts Affected Since May

While Discord confirmed that roughly 200 accounts were erroneously banned over the most recent weekend alone, the company estimates the bug may have impacted approximately 8,200 accounts since the problem first surfaced in May 2026.

The company uses automated, AI-driven content moderation systems that compare uploaded media against known harmful material. Discord’s Trust & Safety team acknowledged that false positives are a known limitation of this approach, and ordinarily, flagged content is reviewed by a human moderator before any enforcement action is taken against the user.

A Double Bug: Wrongful Bans That Couldn’t Be Undone

Discord explained that the intended system behavior is to temporarily pause uploads during the human review period, not to ban the account outright. However, a bug caused accounts to be banned instead. Even worse, when the Trust & Safety team reviewed the flagged content and cleared it as harmless, a second bug prevented the ban from being automatically lifted, leaving affected users locked out of their accounts.

“The intended behavior is to temporarily pause uploads during that review, not ban the account. We had a bug that caused the latter,” Discord’s support team wrote on X. “When our staff reviewed and cleared those accounts, the same bug prevented the ban from being lifted automatically, so it just stayed in place.”

What Kind of Content Triggered the Bug?

The bug’s trigger was surprisingly broad. Any image featuring a square grid pattern could set off the false positive:

Spreadsheets from productivity software
Chessboards showing game positions
Minecraft inventory screens from the popular game
Other grid-based images

This wide trigger range meant that users sharing innocuous gaming screenshots, productivity content, or hobby-related images all risked being banned.

Discord’s Response and Apology

Discord has apologized for the incident, calling it an “embarrassing mistake” to have remained active for approximately two months. The company stated that all accounts wrongfully banned due to the bug should now be reinstated.

The incident highlights the ongoing challenges platforms face when relying on AI-powered moderation tools. While automated systems are necessary to handle the enormous volume of content uploaded to platforms like Discord every day, false positives remain a persistent problem — and when those false positives are compounded by software bugs, the consequences for users can be severe.

Lessons for the Industry

This episode serves as a cautionary tale for the broader tech industry, where AI moderation is increasingly relied upon to police content at scale. Discord’s experience underscores the importance of building robust fallback mechanisms, ensuring that human review processes work as intended, and thoroughly testing automated enforcement systems before deploying them widely.