Community Guidelines & Acceptable Use

Last updated: 14 August 2026

These are the rules for using Commentrs. They apply to everything you post, send, stream or automate — videos, photo carousels, sounds, comments, direct messages, stories, live broadcasts, profiles and API traffic. They are part of the Terms of Service, which is the contract; this page is the same rules written to be read rather than to be litigated. Where the two differ, the Terms govern.

1. How the percentages work

Read this once and every number on the rest of the page means something exact.

• 100% = one strike. A 25% violation is a quarter of a strike, so four of them reach one. There is no separate tally of "warnings"; there is a score.
• Percentages accumulate and do not reset. Forty 2.5% breaches are one strike, arrived at the slow way.
• A clean record earns a discount of up to 25%. Time without a violation builds a good-standing reduction, capped at 25%, and it is applied to the weight of future violations — a 20% harassment finding costs 15% to an account carrying the full 25% discount. It reduces what a NEW violation costs; it never removes one you already have.
• Anything weighted 200% or more ignores the discount entirely. A clean history is not credit against child safety or terrorism, and letting it reduce those would mean a long-standing account got a lower price for the two things that carry no price at all.
• The ladder is by strike, not by violation. Strikes 1 and 2 are a written warning. Strike 3 is a suspension of 1 to 7 days, its length set by history. Strike 4 is a 30-day suspension. Strike 5 is permanent deletion.
• Some rules have no percentage at all, and they are marked. They are the system-abuse rules in section 5: they are enforced by blocking, rate-limiting or removing the account outright, because scoring a script's behaviour against a human ladder is the wrong instrument.

2. The two that end an account immediately

These are weighted 200% each and carry an instant permanent removal — the deletion is arithmetic rather than discretion, and there is no warning ladder in front of it. Where the law requires it, we refer the matter to the relevant authorities.

• Child safety — 200%, immediate permanent deletion. Child sexual abuse material, or any content that harms, sexualises or endangers a child. Reported and referred, always.
• Violent or terrorist content — 200%, immediate permanent deletion. Terrorist propaganda, direct incitement to violence, or content promoting a violent extremist organisation.

3. Serious content violations, and what each costs

These are scored. A single one of the heavier entries here takes most of a strike, and two of the heaviest reach the suspension ladder on their own.

• Illegal goods — 80%. Trading, advertising or facilitating the trade in controlled substances, weapons, or any other unlawful goods or services.
• Malware or phishing — 75%. Viruses, ransomware, credential-harvesting links, or anything designed to compromise another person's device or account.
• Sharing private information (doxxing) — 60%. Publishing somebody's address, phone number, identity documents, workplace, IP address or other personal data without their permission — including assembling it from public sources.
• Hate speech — 50%. Attacking people for who they are: race, ethnicity, national origin, religion, caste, sexual orientation, sex, gender identity, disability or serious disease.
• Scam or fraud — 40%. Fake giveaways, investment and crypto scams, advance-fee schemes, deceptive selling, and fraudulent card details or forged financial documents used against the purchase or withdrawal systems.
• Dangerous acts or self-harm — 30%. Content promoting or instructing self-harm, suicide or eating disorders, and dangerous challenges or stunts presented for imitation.
• Nudity or sexual content — 25%. Sexually explicit material, and gratuitous gore. Where content is mature but permitted, mark it; regional rules still apply to it.
• Copyright infringement — 25%. Posting work you have no right to distribute. Section 8 covers the claim process from both sides.

4. How to treat other people, and what each costs

• Harassment or bullying — 20%. Targeted abuse, pile-ons, threats, sexual harassment and repeated unwanted contact — in comments, in direct messages, or in a live stream's chat.
• Block evasion — 20%, scored as harassment. Using a second account to reach somebody who blocked you is not a separate minor rule; it is the same conduct with an extra step, so it carries the same weight.
• Impersonation — 20%. Pretending to be a specific person or business. Parody and fan accounts are fine when they say what they are, in the name or the bio, where somebody sees it before they are fooled. If an account is running as YOU, see section 8 — an impersonation claim is a different and stronger route than a report.
• Harmful false information — 10%. False claims about health, medicine, elections or an ongoing emergency, presented as fact.
• Profanity banned in your country — 2.5%. The lightest weighting on the platform, and deliberately so: it is regional, it is often accidental, and forty of them are one strike. Comments containing terms blocked for your region, or terms a creator has filtered, are refused as you write them rather than hidden afterwards — because hiding a comment after the fact still sends the notification to the person it was aimed at.

5. Spam, automation and system abuse

The scored rules first.

• Fake likes, views or bot activity — 15%. Views, likes, comments, follows, watch time or donations produced by bots, scripts, macros, click farms or reciprocal-engagement arrangements. Buying or selling any of them is included, and so is being the seller.
• Spam or misleading content — 5%. The same content posted repeatedly; flooding comments or direct messages; bulk unsolicited messages; misleading thumbnails, titles or hashtags; hashtag stuffing or tagging with unrelated trending terms; networks of accounts promoting each other while presenting as unrelated.

The rules below carry no percentage, and that is not leniency. They are enforced by machine, at limits set well above what a person reaches by hand — so a hand is never throttled and a script always is — and the response is a block, a rate limit, suspension or permanent deletion rather than a score. A number here would imply you get several attempts, which is exactly the wrong impression.

• No scraping — no percentage; blocked, then removed. Do not use crawlers, headless browsers or any automated means to extract content, profiles, follower lists or media, and do not build a dataset out of them. Automated access is permitted only through an interface we publish for that purpose, within its documented limits.
• No load abuse — no percentage; blocked, then removed. Automated requests that overload our servers, spamming API endpoints, or conduct resembling a denial-of-service attack.
• No rate-limit evasion — no percentage; the limit simply holds. Circumventing a rate limit, cooldown or per-account cap by any means. Limits apply to likes, live-join requests, invitations, copyright claims (10 per account per day, 3 per browser per day signed out) and support tickets (3 open at a time).
• No reward farming — no percentage; the reward is voided. Creating accounts to abuse the invitation reward, the Diamond system or any promotion. An invitation claimed on the same device as the inviter pays nothing. Idle-watching set-ups that simulate viewing to accrue Diamonds are included.
• No ranking manipulation — no percentage; removed. Of the For You feed, hashtags, trending sounds or search.
• No VPN or proxy use to evade a regional restriction — no percentage; the account is archived until the region is confirmed. See Chapter 2 of the Terms for how a region is set and how an archived account is reactivated.
• No reverse engineering or unauthorised access — no percentage; immediate deletion. Of our interfaces, our systems, or another user's account. Attempting to take over an account is in this line and not in section 4.
• No account sales outside the official feature — no percentage; the sale is not recognised. Transfers arranged elsewhere are not honoured and we accept no responsibility for money lost to them.

Automated crawling of the public website is additionally governed by robots.txt. Following it is necessary but not sufficient: a crawler that obeys robots.txt and still scrapes user content is still breaking this section.

6. Regional law

Content must comply with the law of every region you publish into. Content lawful in one region may be unlawful in another, and you are responsible for each Content Region you have selected — every account may select 2, and business accounts any number.

Selecting a Content Region submits your content to that region's government. Where the head of state of a territory you selected, or a body acting under their authority, requests removal, we act on it and the content is removed or blocked there. It stays visible everywhere else it is lawful and it remains yours. The reason and the requesting territory are recorded on the content and shown to you. This is section 1.13.1 of the Terms.

7. Reporting

Content, accounts, sounds, comments and live streams can all be reported from the app. A report carries a reason and creates a record a reviewer acts on; reviewers see the reported item itself alongside the report, not just a reference to it.

A report is free and carries no cost to you, whatever the outcome. Reporting in bulk and in bad faith — coordinated mass-reporting of an account you dislike — is harassment and is scored at 20% under section 4.

Do not report copyright infringement as an ordinary report. It has its own route in section 8 and its own evidence requirements; filed as a report it will not remove the content.

8. Someone using your content, and what a wrong claim costs

If your sound, your video or your identity is being used by somebody else, file it from Support → Someone is using my own content. The full process is in the Copyright & DMCA policy. Three things are worth stating here because they are where people get caught out:

• You must be the owner. Not a manager, a label contact or a friend. There is no "acting on behalf of" route.
• Reach differs. A sound claim covers every public video using that sound. A video claim covers that one video. A profile claim covers the whole account. Choosing the wrong one is the commonest reason an honest claim is refused.
• Being wrong costs 0% three times, then 5% each. Your first 3 refused claims cost nothing at all — mistaking a licensed re-upload for a theft is an ordinary error and punishing the first one teaches rights holders not to file. The 4th and every one after it costs 5% of a strike. The same counter covers objections you file against claims on your own videos, because they are the same mistake from the other end.
• That 5% cannot be appealed. It is the only non-appealable consequence on the platform, and it is arithmetic: the penalty counts decisions that were each reviewed and each individually appealable at the time. Appeal the refusal, not the total. The count and how many free ones you have left are shown on the form before you swear the statement.
• Bad faith is separate and heavier. A claim that was not merely wrong but dishonest is 40% as fraud under section 3, and a pattern of them is treated as system abuse under section 5.

9. What happens when a rule is broken

Your current standing, every strike on the account and what each one was for are shown at Settings → Account standing — including the percentage that was applied and whether a good-standing discount reduced it.

• Strikes 1 and 2: a written warning.
• Strike 3: suspension for 1 to 7 days, the length set by breach history.
• Strike 4: suspension for 30 days.
• Strike 5: permanent deletion.

A moderation notice is a record. Turning off its notification stops the alert; it does not remove the notice or its effect. Nothing in your settings can make a notice disappear, because "I was never told" has to stay false.

10. Appeals

Appeal from the notice itself or from Settings → Account standing.

• An appeal that succeeds costs 0% and refunds the original. The strike is reversed and its full percentage comes back off your score.
• An appeal found to be plainly unfounded costs a further 5%, and a pattern of them may itself be treated as system abuse under section 5. Where local law does not permit a penalty for exercising a right of appeal, none is applied.
• The repeated-false-claim penalty in section 8 cannot be appealed — the one exception on this page, for the reason given there.

11. Changes

These guidelines change as the platform and the law do. Material changes — including any change to a percentage on this page — are announced with at least 3 days' notice, on the same terms as the Terms of Service.