ToS Phrase Checker
Paste panel text, a bio, or community rules. Flags phrasing that commonly triggers automated moderation or Partner-review rejections.
Client-side only — your text never leaves your browser.
How the checker works
Twitch enforces three overlapping policy layers. AutoMod is the in-chat filter every channel configures: four categories (Discrimination, Sexual content, Hostility, Profanity), each set between level 0 (off) and 4 (strictest). Community Guidelines are the platform-wide rules that apply regardless of AutoMod tier. Content Classification Labels (CCLs) are six self-applied labels that gate audience reach for content like gambling, drugs, or mature games. The scanner above checks your copy against all three layers simultaneously and reports each match with its source citation.
The four AutoMod categories Twitch uses
AutoMod's official taxonomy is fixed: Discrimination & Slurs (race, ethnicity, religion, gender, sexual orientation, disability), Sexually Explicit Language (anatomy, suggestive phrasing, explicit terms), Hostility (insults, threats, mockery, brigading language), and Profanity (general crude language). Each category runs at its own level. On Twitch's 0-4 scale, higher numbers are stricter: level 0 = off (no filtering), level 1 = least filtering (catches the most overt slurs and explicit text only), level 2 = default (catches most strong profanity and clear hostility), level 3 = aggressive (holds category-adjacent words), level 4 = strictest (the broadest filter, appropriate when you want even soft references held). Use the slider above to test "would AutoMod at level X have held this?" before publishing the copy.
The six Content Classification Labels (CCLs)
Twitch publishes exactly six CCLs and the streamer is responsible for selecting them when starting a broadcast: Drugs, Intoxication, or Excessive Tobacco Use; Violent and Graphic Depictions; Mature-Rated Game; Significant Profanity or Vulgarity; Sexual Themes; and Gambling. The Gambling label also requires compliance with the 2022 Gambling Policy and the approved-site list. Slots, casino, and sportsbook content from non-approved operators is removed regardless of label state. The Mature-Rated Game label is applied automatically by Twitch for ESRB-Mature titles, but verify it on stream-start because the auto-application has edge cases. Failing to apply a relevant CCL is one of the more common quiet soft-fails: the channel is suppressed in recommendations without an obvious notification.
Partner review: the adjacent rejections
Most Partner-review rejections that surprise creators come from the adjacent issues, not the obvious violations. A "free subs" giveaway phrased without opt-in language reads as misleading. A title naming a Top-40 song surfaces in DMCA scans even when the audio itself is muted. A bio offering medical or legal "advice" pings the Civic Integrity (Twitch added 2020) and Health Misinformation (added 2021) policy axes in Twitch's Community Guidelines. The scanner flags each of these by category so you can rephrase before re-applying. Pair this scan with the Affiliate Safety Checker for a configuration-side review of strikes, DMCA history, schedule, and moderation posture. The two tools cover different surfaces and reviewers weight both.
Copyrighted music: the DMCA shadow
DMCA takedowns are issued by rights-holders, not Twitch. The platform forwards them. They apply primarily to VODs, clips, and highlights, not the live broadcast. Naming a copyrighted track in a stream title doesn't directly cause a strike, but it gives a rights-holder's legal team a clean attribution surface once the VOD is indexed. Use stream-safe music libraries (Twitch Soundtrack, Monstercat, Pretzel, Epidemic). For unlicensed audio sessions, disable VOD saving so there is no surface to strike, and run a Channel Audit on your back catalogue before applying for any tier promotion.
