You made your first ad with an AI video tool: a smiling woman holds your serum and says it changed her skin. It looks great. Then the worry arrives. Will Instagram stamp "fake" across it? Will TikTok reject it? Could a customer complain to someone in Ottawa? Those are three different questions, and each platform answers the first two differently.
The short answer for a Canadian small business: TikTok's ad policy requires you to disclose significantly AI generated or edited content, with its AI label or your own clear disclaimer, and rejects or restricts ads that are not disclosed. Meta and Google mostly label AI ads themselves, in an "About this ad" or "How this ad was made" panel, and Google added an AI label setting for advertisers in July 2026. Canada has no general AI label law for ads, but the Competition Act already makes any misleading ad illegal, whoever or whatever made it, with civil penalties up to $10 million for a company's first violation. The woman in that ad is the real problem: she is not a customer.
Key takeaways
- TikTok: label significant AI content (fully AI generated video, image or audio, people shown doing things they did not do, cloned voices) with the AIGC label or your own clear disclaimer, or the ad is rejected or restricted.
- Meta and Google: both apply AI labels themselves for some ads, in an ad info panel rather than across the video. Google's AI label setting for advertisers rolled out from July 2026.
- Canada: the Competition Act's general impression test applies to AI ads like any other. An AI "customer" giving a testimonial, or AI footage that improves the product, can be misleading even if every word is true.
This is a practical guide, not legal advice. Platform policies change often; confirm against each platform's policy page before a large campaign.

The rules at a glance
| Platform | What it says about AI in ads | What you do | Where people see it |
|---|---|---|---|
| TikTok | Significantly edited or AI generated content is allowed only if disclosed; undisclosed AI content gets the ad rejected or restricted | Apply TikTok's AIGC label, or add your own clear disclaimer, caption, watermark or sticker | On the ad |
| Meta (Facebook, Instagram) | Labels ads made or significantly edited with its own generative AI tools; from June 2026 also detects third party AI tools through industry standard signals | For ordinary ads, nothing extra is described; political and social issue ads must disclose digital creation or alteration | "AI info" inside "About this ad" in the three dot menu |
| Google Ads | AI label setting for image and video creatives generated or modified with AI, rolled out from July 2026; some labels applied automatically to assets from Google's tools | Use the AI label setting or put a label in the creative; election ads tick "Altered or synthetic content" | "How this ad was made" in the ad's menu, in some countries |
| YouTube (organic uploads) | Creators must disclose realistic altered or synthetic content | Select "Yes" under AI use when uploading in YouTube Studio | A disclosure on the video |
TikTok: the strictest of the four
TikTok's ad policy on misleading and false content is the clearest. Significantly edited or AI generated content is allowed if you apply TikTok's AIGC label or add your own clear disclaimer, caption, watermark or sticker. If TikTok finds AI generated content that was not disclosed, the policy says the ad will be rejected or restricted.
What TikTok counts as a significant edit:
- Fully AI generated images, video or audio.
- A primary subject shown doing something they did not do.
- AI voice cloning that makes someone say something they did not say.
What it does not count: adjusting lighting, brightness or saturation, removing or changing a background, and denoising.
Two other rules in the same policy catch AI ads often. Ads must not promise or exaggerate a product's effect, and before and after product effect comparisons that could give a false impression are not allowed. TikTok also prohibits using a public figure's likeness without permission, particularly in misleading contexts such as a false endorsement, and says it removes content using a private person's likeness without permission.
Meta: labels it applies itself
Meta's newsroom describes how its ad labels work. Ads whose images or video were created or significantly edited with Meta's own generative AI creative features get an "AI info" label. If Meta's tools made no significant edit and the ad has no photorealistic human, no label applies. In its June 1, 2026 update, Meta said it would also detect ads made with third party AI tools using industry standard signals and label those, and that the label now sits in "About this ad", a destination in the three dot menu on every ad. Meta notes this can vary by region because of local legal requirements.
For organic posts, Meta has used the "AI info" label since July 2024, applied when it detects industry standard AI indicators or when people disclose AI content. Separately, since January 2024, advertisers must disclose when they digitally create or alter certain political or social issue ads, and ads with content that fact checkers have debunked are rejected.
The articles we reviewed do not describe a general disclosure requirement for ordinary commercial ads. That does not make a misleading AI ad acceptable on Meta; it means the label question is mostly handled by Meta, and the honesty question is yours.
Google Ads and YouTube
Google's advertising policy help says that starting July 2026 it rolled out an AI label setting across Google Ads, Display and Video 360, Campaign Manager 360, Merchant Center and Ads Editor, for image and video creatives generated or modified with AI. Advertisers can also put text or visual AI labels inside the creative, and Google says those will not break its text overlay or watermark policies. Google cites AI rules in the European Union, India and New York as the reason, and warns that using the setting does not by itself guarantee compliance with any regulation.
On the viewer side, Google's My Ad Center help describes a "How this ad was made" option in the ad's menu, shown for ads with AI edited or AI created assets, available in some countries. Google says it adds SynthID and C2PA metadata to images and videos generated with its own ad tools. Election ads with consequential synthetic or altered content need the "Altered or synthetic content" checkbox.
For organic YouTube videos, creators must disclose realistic altered or synthetic content at upload, such as making a real person appear to say something they did not, or generating a realistic scene that did not happen. Beauty filters, colour fixes, captions and AI help with scripts or thumbnails do not need disclosure, according to YouTube's help page.
What Canadian law adds
We found no general Canadian federal rule requiring a label on every AI generated ad. What applies to every ad is the Competition Act. The Competition Bureau explains that it prohibits promoting a product or business with claims that are false or misleading in a material respect, meaning anything that could influence whether someone buys. Nobody needs to have been fooled for a claim to break the law.
The test that matters most for AI ads is the general impression test: courts look at the overall impression an ad gives, not only its literal words. The Bureau gives examples that fit AI ads uncomfortably well, including accurate text paired with visuals that show a different product from the one advertised, and claims that are literally true but leave out information that would change a buyer's decision.
Under the civil provisions, the Bureau lists administrative monetary penalties for a first violation of up to the greater of $750,000 or three times the benefit for an individual, and up to the greater of $10,000,000 or three times the benefit (or 3% of worldwide gross revenue) for a corporation. Knowingly or recklessly misleading the public is a criminal offence.
Where AI ads get small businesses into trouble:
- Fake testimonials. An AI person saying "I used this for three weeks and my skin cleared up" presents an experience nobody had.
- A better product than the real one. Bigger portions, shinier fabric, smoother results than a customer will get.
- Implied claims. Health effects, "made in Canada", a discount: anything the video implies must be true.
- Staff or "customers" with undisclosed ties. The Bureau has said employees reviewing their employer must disclose the connection; an AI version of a customer has no connection to disclose because the customer does not exist.

What usually does not need a label
Across TikTok's ad policy and YouTube's disclosure rules, the line is realism: content a viewer could mistake for a real person, place or event. Basic edits such as lighting, colour, background removal and denoising are named as not significant by TikTok, and YouTube lists AI help with scripts, titles and thumbnails as production assistance that does not need disclosure. Writing your caption with an AI assistant is not what these rules target. Check each platform's page for your exact case, because the rules are not identical.
Checklist before you upload an AI ad
- Does it show a realistic AI person, voice or place? On TikTok, add the AIGC label or your own clear disclaimer. On Google, use the AI label setting. On Meta, expect an AI info label in "About this ad".
- Does any AI person appear to be a real customer? Rewrite it. Show the product in use instead, or use a real customer who agreed to appear.
- Does the product on screen match what arrives? Size, colour, texture, portion. Compare frame by frame with a real photo.
- Is every claim true and provable? In the video, captions, voiceover and landing page.
- Is there a before and after? On TikTok, product effect before and afters are not allowed. Everywhere, they must be real.
- Political or social issue content? Stricter disclosure rules apply on Meta and Google. Check before spending.
- Keep your files. Save the prompts, source images and brief. If an ad is ever questioned, you can show how it was made.
For help writing a brief that keeps AI output honest, see how to brief an AI video ad.
Frequently asked questions
Do I have to label AI generated ads on TikTok?
Yes, when the content is significantly AI generated or edited, such as fully AI video, a person shown doing something they did not do, or a cloned voice. Use TikTok's AIGC label or your own clear disclaimer; undisclosed AI content can get the ad rejected or restricted.
Will Meta put an AI label on my Instagram ad?
Possibly. Meta labels ads created or significantly edited with its own generative AI tools, and from June 2026 also ads it detects were made with third party AI tools. The "AI info" label appears in "About this ad" in the three dot menu.
Is there a Canadian law that requires AI ads to be labelled?
We found no general federal rule requiring a label on every AI ad. The Competition Act still prohibits false or misleading advertising, judged on the general impression an ad gives, so AI ads must represent the product accurately and must not present invented customers as real.
What does Google Ads require for AI ads?
Google introduced an AI label setting for image and video ads generated or modified with AI from July 2026, and allows labels inside the creative. Election ads with consequential synthetic content must tick the "Altered or synthetic content" checkbox.
Sources
- TikTok Ads Policy: Misleading and false content
- Meta: Expanding GenAI transparency for Meta's ads products
- Meta: Our approach to labeling AI generated content and manipulated media
- Google Advertising Policies Help: Updates to AI labeling requirements (July 2026)
- Google My Ad Center Help: AI transparency in ads
- YouTube Help: Disclosing use of GenAI content
- Competition Bureau Canada: The general impression test
- Competition Bureau Canada: False or misleading representations
- Competition Bureau Canada: Online reviews posted by employees, businesses could be liable



