YouTube, TikTok and Instagram Misinformation Rules: A Plain-English Guide for Creators

What is YouTube's misinformation policy?

YouTube's misinformation policies don't ban being wrong. As of September 2026, they ban specific kinds of false content that can cause serious harm, with separate pages for elections and medical misinformation. The general policy covers three categories:

• Census suppression: misleading people about the time, place, means or eligibility for the census, or false claims that could materially discourage people from taking part.
• Manipulated content: video technically doctored in a way that misleads and may pose a serious risk of egregious harm. YouTube notes this usually goes beyond clips simply taken out of context.
• Misattributed content: old footage from a past event passed off as current, where that may pose a serious risk of egregious harm.

Why roast and debunk channels should care: you replay false claims for a living, and YouTube's exceptions are your lane.

YouTube may allow content that would otherwise break these rules if it adds educational, documentary, scientific or artistic (EDSA) context in the video, audio, title or description. It may also make exceptions for public-interest content and for content whose purpose is to "condemn, dispute, or satirize" misinformation, while saying the exception isn't a pass to promote misinformation. It's still YouTube's call, not something you earn by typing "satire" in the description.

Example: play the clip ("this tea cures diabetes"), freeze the frame, and put up on-screen text like "CLAIM: tea cures diabetes. What health authorities say: (your one-line summary). Source in description." Then land your joke. Use the same framing in the title (for example "Debunking the tea that 'cures' diabetes") and list your sources in the description.

YouTube medical and election misinformation policies

Health and elections are where debunk channels get burned, because the claim you're mocking is often exactly what the policy bans.

Medical. The medical misinformation policy covers content that contradicts local health authority guidance on preventing or transmitting specific health conditions, on the safety, efficacy or ingredients of approved and administered vaccines, and on treatments for specific conditions. Exceptions exist for EDSA context, for content that condemns, disputes or satirizes the misinformation, and for personal experiences. The policy can change as health-authority guidance changes, so re-read it before each health video.

Elections. The elections misinformation policy targets voter suppression (misleading people about when, where or how to vote, or who's eligible), false claims about the technical eligibility of current candidates and sitting officials (such as age, citizenship or whether they're alive), and encouraging others to interfere with democratic processes. The same EDSA and public-interest exceptions apply.

Money is a separate question. Under YouTube's advertiser-friendly guidelines, content that spreads misinformation denying established medical or scientific facts, or promotes scientifically unproven remedies, earns no ad revenue. That's judged separately from the Community Guidelines, so a video can stay up and still earn nothing.

How does a YouTube misinformation strike work?

There's no separate "misinformation strike." A misinformation violation runs through YouTube's standard Community Guidelines strike system. As of September 2026:

• First violation: usually a warning. Complete the policy training and the warning can expire after 90 days. Break the same policy again inside that window and it may not.
• Strike one: a one-week freeze on uploading, live streaming and other posting. The strike stays on your channel for 90 days.
• Strike two within 90 days: a two-week freeze.
• Strike three within the same 90 days: may lead to permanent channel removal.
• Severe abuse: a single case can end a channel with no warning at all.

You can appeal a strike if you think YouTube got it wrong. Copyright strikes are separate from Community Guidelines strikes, and YouTube says a Content ID claim typically doesn't lead to a copyright strike. Our guide to reaction videos and fair use covers that side.

TikTok misinformation policy: removals, the For You feed and strikes

TikTok's rules live in its Integrity and Authenticity guidelines. The current version was released August 25, 2026 and took effect September 24, 2026. Removed, whatever the poster's intent: misinformation that could cause significant harm, such as:

• False claims that risk public safety or cause panic, including a past crisis passed off as recent.
• Dangerous health misinformation, like promoting unproven treatments that may be fatal or discouraging professional care for life-threatening conditions.
• Denying climate change or misrepresenting its causes.
• Harmful conspiracy theories and hoaxes.
• Misinformation about how, when and where to vote, voter eligibility, election procedures or results.

Kept out of the For You feed:

• Misinformation that may cause moderate harm.
• Out-of-context footage that may mislead on a topic of public importance.
• Misuse of authoritative sources to push misleading conclusions.
• Unverified claims about a crisis or major civic event, and anything under review by TikTok's independent fact-checkers.

TikTok may also add warning labels or prompt people to pause before sharing unconfirmed content.

Explicitly allowed: personal opinions without harmful misinformation, personal medical experiences that don't discourage professional care, and documentary or educational content reporting on or condemning misinformation. Make that purpose obvious inside the video (on-screen text, caption or voiceover), because an allowed video can still lose For You reach or get a warning label.

Strikes and bans. A 2023 TikTok explainer said an account gets a strike when violating content is removed, hitting the strike threshold for a feature or policy means a permanent ban, severe violations can bring an immediate ban, and strikes expire after 90 days. The thresholds may have changed since.

Stitching a claim to correct it? Read how to check a TikTok before you stitch it first.

Instagram false information: Community Notes vs fact-checkers

Instagram, like Facebook and Threads, runs two systems depending on location, as of September 2026:

• In the US: Meta announced on January 7, 2025 that it would replace third-party fact-checking with Community Notes, and says it stopped showing new fact-check labels on US content from April 7, 2025. Meta says notes add context but don't affect who sees a post or how widely it's shared.
• Outside the US: IFCN- or EFCSN-certified fact-checkers still rate content False, Altered, Partly False, Missing Context, Satire or True.
• What the ratings do: False and Altered posts get distribution dramatically reduced plus a context notice; Partly False gets a smaller cut; Satire and True carry no label or restriction.
• Latin America test: as of September 2026, Meta says it's beginning to test Community Notes in 16 Latin American countries. Notes aren't published there yet, and Meta says it will move away from fact-checking in those countries only once it's confident the product works (Meta Transparency Center).
• Repeat posting outside the US: accounts that repeatedly post content rated False or Altered face penalties lasting 90 days, such as lower placement in Instagram Feed and Stories and reduced recommendations. In the US, Meta says it stopped demoting fact-checked content and applying strikes for sharing it.

Some misinformation is removed everywhere, rated or not. Meta's Misinformation Community Standard removes content likely to directly contribute to imminent violence or physical harm, harmful health misinformation such as certain false vaccine claims and dangerous fake cures, and voter or census interference.

Do you have to label AI or edited clips?

If your roast uses AI voices, AI visuals or aggressive cuts of someone's words, this is where you can get flagged even when every claim you make is right. As of September 2026:

• YouTube, when to disclose: you must disclose AI that meaningfully alters or generates photorealistic content, such as making a real person appear to say or do something they didn't, altering footage of real events or places, or generating realistic scenes that never happened.
• YouTube, when you don't have to: clearly unrealistic content, minor edits like color fixes or beauty filters, and AI help with a script, title or thumbnail.
• YouTube labels: YouTube's help page says photorealistic AI content can get a label in the video player itself, and that disclosing won't limit a video's audience or its eligibility to earn money. (In a March 2024 post, YouTube also said videos on sensitive topics like health, news, elections or finance get a more prominent label.)
• YouTube penalties: YouTube may add a label itself, and creators who consistently don't disclose can face content removal or suspension from the YouTube Partner Program.
• TikTok, when to label: you must label AI-generated or significantly edited content that shows realistic scenes or people, with the AIGC label or your own clear caption, sticker or watermark.
• TikTok, what counts as significantly edited: media that makes it seem someone did or said something they didn't (or changes their appearance until they're unrecognizable), for example by cutting phrases to change their meaning, rearranging or combining clips, changing speed, or adding or removing audio or video. Small edits like color correction, reframing or cropping don't need a label.
• TikTok, AI voices: AI audio that mimics a real person's voice needs disclosure; generic text-to-speech that isn't a recognizable voice of a known person doesn't. TikTok said in May 2024 that it auto-labels some AI content made elsewhere by reading C2PA Content Credentials.
• Instagram, disclosure: Meta said in February 2024 it would require its disclosure tool for photorealistic video or realistic-sounding audio that was digitally created or altered, may penalize people who skip it, and reads C2PA and IPTC metadata.
• Instagram, labels: Meta's AI info label can be made more prominent when digitally created or altered content creates a particularly high risk of materially deceiving the public on a matter of importance.

How to avoid a misinformation strike: pre-post checklist for roast and debunk videos

Run this before any upload that repeats someone else's claims. It won't make a video strike-proof, and enforcement is always the platform's call, but it targets the mistakes the policies above describe and the ones that get creators debunked back.

1. 1

   Separate claims from jokes

   List every factual statement in your script: numbers, dates, "studies show," who said what. Jokes and opinions are fine; stated facts are what get you corrected or flagged.

2. 2

   Get a receipt for every fact you state

   Find a primary or authoritative source for each. For health topics, check current health-authority guidance, the yardstick YouTube's medical policy uses.

3. 3

   Quote your target exactly

   Use their exact words with timestamps, not a paraphrase from memory. A misquote hands your target the easiest comeback there is. A timestamped transcript makes this quick, but check each quote against the video itself, since auto-captions can mishear words.

4. 4

   Put the correction next to the claim

   Never play a harmful claim without the dispute right beside it, in on-screen text, voiceover or both. That makes the dispute purpose that YouTube's and TikTok's exceptions talk about obvious, even to someone who only sees a clip.

5. 5

   Show your sources

   Put your sources in the description, and in a pinned comment if you use one. Anyone trying to debunk you back should run into your receipts first.

6. 6

   Check your cuts and your dates

   Trimmed or rearranged someone's words? Make sure the edit doesn't change their meaning. On TikTok, a realistic edit that changes what someone appears to say needs a label, and even with a label it isn't allowed if it misleads about a matter of public importance, such as what a public figure said. Put the original date on old clips so they can't read as current.

7. 7

   Label AI and heavy edits

   Use each platform's disclosure where it applies. It's your job even if the platform might auto-detect it.

8. 8

   Make the title and thumbnail honest

   YouTube's spam and deceptive practices policy covers maliciously misleading titles and thumbnails. Outrage is fine; bait-and-switch isn't.

9. 9

   Roast the claim, not the body

   YouTube's harassment policy bans prolonged insults based on physical traits or protected group status, and its comedy exception isn't a pass to harass. TikTok bans insulting physical attributes and protects private figures more than public ones. Go after what people said and did. BSmeter's Roast mode works the same way: it goes after the claims, not people's looks. More in how to make roast videos.

What BSmeter can (and can't) do for you here

BSmeter handles the research half: finding the shakiest lines in a video before they end up in yours, and searching for receipts on the weakest ones, so your roast is harder to debunk back.

Paste a YouTube, Shorts, TikTok, Instagram Reel or podcast link at bsmeter.ai with a free account. It transcribes the video, pulls out the checkable claims with timestamps, scores each one 0-100 for credibility, names the framing move (such as Missing Context, Confidence Mismatch, Anonymous Authority or Cherry-Picked), flags logical fallacies in Analyze mode, and searches for evidence sources on the claims that score low.

Run it on the video you're roasting to find its weakest claims. Each claim comes with its timestamp, and the low scorers come with any evidence sources BSmeter found. That's raw material for the on-screen correction in step 4, but open and check each source yourself before you cite it. Regular analyses are public pages, so don't paste anything you haven't published yet.

Free accounts get 3 analyses a day on videos up to 30 minutes (see plans). For the full workflow, read how to debunk a viral video, or check a TikTok's claims or an Instagram Reel's claims.