Discord的AI因棋盤圖案誤封8000+用戶
discord恢復了8000多名用戶，此前一個bug將電子表格等無害圖像錯誤標記為兒童安全材料。這凸顯了自動審核的風險。
discord fixed a bug in its automated moderation system that wrongfully banned more than 8,000 users over two months after flagging benign images such as game textures and grid patterns as illegal content. the glitch bypassed the standard human review step, causing immediate account suspensions for content that vaguely resembled known harmful material in databases. discord stated it is implementing safeguards to prevent recurrence, but the incident sparked outrage among users who rely on the platform for livelihood and community. this case signals a growing challenge for tech companies relying on ai to police content at scale, proving that false positives in automated safety systems can have severe consequences for user trust.