# AI Moderation

> Moderation that judges intent and context rather than matching keywords — catching harassment phrased without a banned word, and ignoring a banned word used harmlessly.

- Canonical: https://discordtools.blog/glossary/ai-moderation
- Group: Moderation
- See also: https://discordtools.blog/glossary/automod, https://discordtools.blog/glossary/moderation-case
- Updated: 2026-08-31
- Publication: The Index, an independent site not affiliated with Discord Inc.

---
Keyword filters fail in both directions, and both failures are expensive: they miss the abuse
that avoids the word list, and they punish members quoting or discussing a term innocently.
Context is the whole point.

The operational questions are what happens on a false positive, whether a human reviews before
action, and whether decisions are logged with a reason a moderator can read.

**Vetox** ships AI Auto-Moderation and AI Server Control on the Black tier, with a monthly
token allowance.