Reddit Brings LLMs to Community Moderation

Reddit’s communities largely rely on volunteer moderators to write and enforce their own rules. Its longstanding AutoModerator system uses keywords, regular expressions and account attributes, making it powerful but often difficult to configure and less adept at interpreting context. Rules Hub instead uses large language models to assess whether posts and comments violate the intent of a rule, aiming to reduce routine work involving spam, abuse and generic violations while keeping human moderators in control.

Reddit began testing Rules Hub with a small group of communities on April 16, 2026, and announced a broader rollout on August 5. A day later, access expanded to every community created since July 5 and all newly created communities going forward. Moderators can test rules against historical content, then choose whether matching posts or comments should be removed, reported or only logged. Existing communities are unchanged for now, with wider availability planned for later in 2026.