Key facts
- Discord's AI moderation system mistakenly banned over 8,000 users.
- The bans occurred over a two-month period.
- The AI flagged harmless images, including spreadsheets and transparent backgrounds.
- Discord is restoring affected accounts.
- The company is working on improved safeguards for its AI moderation.
Discord has identified and is rectifying an artificial intelligence moderation bug that led to the wrongful banning of over 8,000 users within a two-month period. The AI system incorrectly flagged and banned users for uploading seemingly harmless images. Reports indicate that the types of images that triggered these erroneous bans included spreadsheets, images with transparent backgrounds, and other content that posed no violation of Discord's terms of service. The company has acknowledged the error and is actively working to restore the accounts of all mistakenly banned users. In response to this incident, Discord is also focusing on enhancing its AI moderation safeguards to prevent similar occurrences in the future. The company aims to improve the accuracy of its automated systems to better distinguish between policy-violating content and benign uploads.