Quick answer: Using scraped web content to train an AI model carries real legal risk in Canada, and the risk does not come only from copyright. Website terms of use, privacy law and contract claims all apply. Two cases show the range: Canadian news publishers are suing OpenAI in Ontario over copyright, and Reddit sued Anthropic in the U.S. over breach of its user agreement. If you build AI products, know where your training data comes from. If you run a platform, your own terms of service are your first line of defence.

Key takeaways: Document the source and licence of every dataset, read the terms of the sites you collect from, keep personal information out of training data unless you have a lawful basis, and update your own terms of service to say clearly whether and how others may use your content for AI training.

What is the Canadian lawsuit against OpenAI about?

A group of Canadian news organizations sued OpenAI in Ontario, alleging that their articles were copied without permission to train ChatGPT. OpenAI challenged whether an Ontario court could hear the case. In November 2025 the court held that it has jurisdiction, so the claims will proceed in Canada. The court has not decided whether training on the articles infringes copyright. That question is still open.

What is Reddit v. Anthropic about?

Reddit sued Anthropic in California, alleging that its content was collected and used for AI training in breach of Reddit's user agreement and terms. The case was sent back to California state court on March 30, 2026. It matters for founders because the claim is built on contract and terms of use rather than copyright. A platform does not have to own the copyright in user posts to have a claim if its terms restrict how the content can be collected.

Is web scraping legal in Canada?

There is no single rule. Collecting publicly available information is not automatically unlawful, but several laws can apply at once:

• Copyright: copying protected works can infringe unless an exception, such as fair dealing, applies. How fair dealing applies to AI training has not been settled by a Canadian court.
• Contract: if a site's terms of use prohibit scraping and you are bound by them, scraping can be a breach of contract.
• Privacy: scraped content often includes personal information. Canada's privacy commissioners have said publicly accessible personal information is not free to use for any purpose under PIPEDA. Québec's Law 25 adds its own requirements.

What should AI startups do about training data?

• Keep a data inventory: where each dataset came from, under what licence or terms, and when.
• Prefer licensed, open-licensed or first-party data, and keep the licences.
• Respect robots.txt and site terms, and stop collection when asked.
• Filter personal information out of training data or confirm you have a lawful basis for it.
• Check what your model providers promise. Indemnities and data-use terms in vendor contracts differ widely.
• Expect enterprise customers to ask about this in security reviews and SaaS agreements, and have honest answers ready.

What should platforms and SaaS companies put in their terms?

If you host content, including user-generated content, your terms of service should address AI directly: whether automated collection is allowed, whether content may be used to train models, what licence you take from users, and whether you use customer data to improve your own AI features. That last point belongs in your privacy policy too. Vague terms make enforcement harder and erode customer trust.

FAQ

Can I train an AI model on public websites in Canada?

Not freely. Public availability does not remove copyright, contract or privacy obligations. The risk depends on what you collect, from where, and what you do with it.

Does fair dealing cover AI training in Canada?

No Canadian court has decided that yet. The federal government has consulted on copyright and generative AI but has not changed the law.

Can my terms of service stop others from scraping my site?

Clear terms help, especially when users have to agree to them, as Reddit's case shows. They work best alongside technical measures and a consistent enforcement record.

Building an AI product or protecting your platform's content? For Founders Law drafts AI terms of use and acceptable use policies, terms of service and privacy policies, and reviews model provider and data licence agreements for a flat fee agreed before work starts. Book a free consultation.

By Brooke Ash, BA, JD, founder of For Founders Law PC, licensed by the Law Society of Ontario. Last reviewed October 27, 2026. General information only, not legal advice. Related reading: AI and privacy law tracker, privacy lessons from the Meta smart glasses backlash and our startup, AI and tech law glossary.