Everything your chat needs to stay welcoming.

Fast rules, thoughtful decisions

Detect flooding, repeated and similar messages, unwanted links, excessive capitalization, and potential personal information. Exempt trusted users and adjust rules for your community.

Local AI, clear boundaries

A local multilingual toxicity classifier adds another signal. Scores are uncalibrated model outputs, and AI findings default to human review. Deterministic rules keep working when AI is unavailable.

One place to stay in control

Review activity and moderation logs, configure escalation, pause moderation, and inspect expiring User Moderation Profiles where permitted. Permanent bans are disabled by default.