AI Creator Rules in Canada (2026): Disclosure, Deepfakes, Online Harms
Canada's rules for AI creators in 2026: Competition Bureau disclosure, Ad Standards' AI influencer guidance, the new deepfake offence, Bill C-34 and CASL.
The Short Version
As of October 19, 2026, no Canadian law makes creators label AI-generated content. What applies instead is law that already reaches AI: advertising rules that require paid relationships to be disclosed and ban misleading impressions, a Criminal Code offence that since July 18, 2026 covers sexual deepfakes of real people, provincial intimate-image and privacy laws, and anti-spam rules for promotional messages. Parliament is also debating a new online harms bill that would put labelling and takedown duties on platforms.
For an AI persona, that comes down to four habits: disclose every brand deal and affiliate link, say the persona is virtual, never use a real person's likeness without written consent, and keep every character unmistakably adult. New to the business model? Start with What Is AI OFM?
Advertising: What the Competition Bureau Expects
The Competition Act's deceptive marketing rules apply to influencers the same way they apply to any advertiser, the Competition Bureau says on its influencer marketing page. Disclose every material connection: payment or commission, free products or services, discounts, trips and event tickets, and family or business ties. The disclosure has to be prominent, in each post, visible without tapping to expand it, and repeated on every platform where the content appears. A brand tag, a discount code or an affiliate link on its own is unlikely to be enough.
The penalties apply to individuals too. Under section 74.1, an individual can be ordered to pay up to CA$750,000 (CA$1 million for each later order), or three times the benefit if that's more. Since June 20, 2025, private parties can also ask the Competition Tribunal for leave to bring deceptive marketing cases, not only the Bureau (Competition Bureau).
The Bureau hasn't published guidance on AI influencers. Its September 23, 2026 submission on AI transparency says the Act is technology-neutral with no exemption for AI, and that labels alone may not be enough to stop AI content from misleading people.
AI Influencers: Ad Standards' 2025 Guidelines
Ad Standards, the advertising industry's self-regulator, added AI to its Influencer Marketing Disclosure Guidelines in the fall 2025 edition. They aren't law, but its Standards Council can consider them when it rules on complaints. The AI points:
An AI influencer still discloses. Being virtual doesn't remove the duty to disclose paid relationships.
Say it's virtual. Brands working with an AI influencer are advised to disclose that the spokesperson is virtual, with tags such as #VirtualInfluencer or #AIinfluencer.
No experiences it can't have. An AI persona shouldn't give testimonials about things it can't experience, such as how a food tastes.
Label AI product images. An image of a product's results or features that is AI-generated or heavily AI-altered should be disclosed, for example with #MadeWithAI, and a disclaimer doesn't fix an image that misleads.
The general rules apply as well: put the disclosure at the start, before any hashtags; #ad is the clearest tag, while #partner, #spon or #collab alone aren't enough; in videos, disclose out loud or on screen within the first 30 seconds; and disclose in the language of the post, so French content gets a French disclosure. More on brand work: How Do AI Influencers Make Money?
Real People and Deepfakes
A crime since July 18, 2026. The Protecting Victims Act (Bill C-16) amended section 162.1 of the Criminal Code, the offence of sharing intimate images without consent. It now also covers images made by electronic means, including AI software, that show an identifiable person nude, nearly nude or in sexual activity and are likely to be mistaken for a real recording of that person. The maximum is 10 years in prison, or 14 in aggravated cases. Threatening to share such an image is a separate offence, and when a court orders an image deleted, it must set a 48-hour deadline.
Provinces add civil routes. British Columbia's Intimate Images Protection Act covers fake, altered and AI-generated images, and its Civil Resolution Tribunal can now award up to CA$75,000 in damages. Manitoba's law has covered fake intimate images since 2024 and was expanded on June 1, 2026. Québec's act on non-consensual sharing of intimate images, in force since 2025, covers altered images too, and anyone other than an individual who ignores an order faces fines of up to CA$50,000 a day.
Privacy. In findings released June 11, 2026 about X and xAI's Grok, the federal Privacy Commissioner held that an AI-generated image of an identifiable person is personal information even though it isn't authentic, and that sexualized deepfakes need express consent. In British Columbia, using someone's likeness to advertise or promote something without consent is a wrong they can sue over under the Privacy Act.
The simple rule is the one in Spicy Studio's Terms: no real, identifiable person without documented consent.
Fictional Personas and the Adult Line
The new deepfake wording is about identifiable people, and a persona invented from scratch shows no one in particular. Section 163.1 works differently. Canada's offence for child sexual abuse and exploitation material covers any visual representation, made electronically or not, of a person shown as under 18 in sexual activity. No real child is needed, and a stated age doesn't change how a character looks. Keep every persona clearly adult in face, body and setting; Spicy Studio's Terms forbid anything else.
Where an adult persona can be posted at all is up to each platform. Is AI OFM Legal? goes through Fanvue, OnlyFans, Fansly, Instagram, TikTok and X.
Online Harms: What Died and What's Pending
The Online Harms Act is dead. Bill C-63, introduced in February 2024, died when Parliament was prorogued on January 6, 2025.
Its successor isn't law yet. Bill C-34, the Safe Social Media Act, had first reading on June 10, 2026 and is at second reading in the House of Commons. As introduced, it would cover larger social media services, expressly including adult content and live-streaming services. They would have to make non-consensual intimate content, deepfakes included, inaccessible within 24 hours of finding it or of a user flagging it, and label synthetic content that could pass for a real recording, under criteria set by regulation. Penalties could reach CA$10 million or 3% of gross global revenue, whichever is greater. The duties fall on platforms, but they would reach creators through each platform's rules.
Age verification. Bill S-209, which would require age verification for online pornography, passed the Senate on April 15, 2026 and is at second reading in the House.
No AI act. The Artificial Intelligence and Data Act died with Bill C-27 in January 2025, and no federal bill has replaced it. A federal consultation on AI transparency, open from July 23 to September 23, 2026, asked how AI-generated content should be identified; its report isn't out yet.
Messages to Fans: CASL
Canada's anti-spam law covers commercial electronic messages sent to an electronic address, which the Act defines as an email, instant messaging or telephone account, or any similar account. Promotional emails count, and so can DMs: a message like "new set is live, 50% off" needs the recipient's consent, has to identify you with contact details, and must offer a free unsubscribe that you honour within 10 business days (section 6 and section 11). The CRTC enforces it, with penalties of up to CA$1 million per violation for an individual.
Posting to Québec
Québec's language law reaches commercial content online. The OQLF's 2025 social media guide says a business's commercial posts aimed at the Québec market, videos, text in images and stories included, must be in French. Other languages can sit alongside if the French version is at least as easy to reach. Whether that catches a creator based outside Québec depends on the facts, so check before running promotions aimed there.
Not Legal Advice
We checked each rule against the official pages linked above on October 19, 2026. This is general information, not legal advice: bills change as they move through Parliament, provinces differ, and your setup may raise questions this page doesn't cover. For anything that matters, ask a lawyer licensed in your province. Buying from Canada? The billing FAQ covers currency, GST/HST and receipts. Working in Australia too? See AI Creator Rules in Australia.
Frequently Asked Questions
Do I have to label AI-generated content in Canada?
+
No Canadian law requires creators to label AI content as of October 19, 2026. But the Competition Act bans misleading impressions, Ad Standards' guidelines advise disclosing that an AI influencer is virtual, and platforms such as TikTok and Instagram have their own AI labels. Bill C-34, not yet law, would make platforms label synthetic content.
Are sexual deepfakes illegal in Canada?
+
Yes. Since July 18, 2026, the Criminal Code offence of sharing intimate images without consent covers AI-made images of an identifiable person that are likely to be mistaken for a real recording, with a maximum of 10 years in prison. Several provinces, including British Columbia, Manitoba and Québec, also let victims get such images removed and claim damages.
Is the Online Harms Act law?
+
No. Bill C-63 died when Parliament was prorogued on January 6, 2025. Its successor, Bill C-34 (the Safe Social Media Act), had first reading on June 10, 2026 and is still at second reading in the House of Commons.
Can an AI influencer promote products in Canada?
+
Yes, with disclosure. Paid relationships must be disclosed clearly in each post, and Ad Standards' 2025 guidelines add that an AI influencer should be identified as virtual and shouldn't give testimonials about experiences it can't have, such as taste.
Do anti-spam rules apply to DMs I send fans?
+
They can. CASL covers commercial messages sent to email, instant messaging or similar accounts. A promotional message needs the recipient's consent, must identify you, and needs an unsubscribe option honoured within 10 business days.
Is a fictional AI persona covered by the deepfake law?
+
The deepfake offence is about images of an identifiable person, and a persona invented from scratch shows no one in particular. A character shown as under 18 in sexual content is illegal under section 163.1 whatever age is claimed, so keep every persona clearly adult.
Want to run this yourself? Every new account starts with free credits.
Explore the AI OFM Track