Discord's automated safety system wrongfully banned more than 8,000 user accounts since May 2026 after a bug caused it to flag harmless images as policy violations, the company confirmed this week.

Around 200 users were banned specifically for posting "grid-like" images — chessboards, game textures, Minecraft inventory screenshots, and similar patterned content that the AI incorrectly matched against known harmful material. An additional 8,000 accounts were caught up for posting what Discord described as "other benign images."

Stanislav Vishnevskiy, Discord's co-founder and CTO, acknowledged the error publicly. The safety system is designed to flag content by matching it against a database of known harmful material using perceptual hashing. When the system produces a false positive — which Discord says is expected to happen — a human moderator is supposed to review and clear the flagged account.

But a second bug compounded the problem: even after staff reviewed and cleared the accounts, the ban was not lifted automatically. "It just stayed in place," Discord stated.

The wave of bans triggered outcry across Reddit and social media over the past week, with users sharing stories of losing access to their communities over completely innocuous images. Discord says all affected accounts have now been restored.

The incident highlights the brittleness of automated content moderation at scale. Pattern-matching systems remain vulnerable to false positives that can silently strip users of access to their digital communities, even when human oversight is nominally in the loop. A review process is only as good as the systems that carry out its decisions.

Sources: The Verge