AI and copyright in Canada: Ottawa still says “stay tuned”

Generative artificial intelligence models learned to write, draw and compose by reading millions of works: books, news articles, songs, images. Most of these works are protected by copyright, and their creators were neither consulted nor paid.

Governments everywhere are looking for an answer. In Canada, the question has been open since 2023, and it still is. In June 2026, asked by Global News about the place of copyright in the new federal AI strategy, the parliamentary secretary to the Minister of AI replied that the issue was still under review and that Canadians should “stay tuned”.

Here is where the file stands, and what it means for creators and rights holders.

A consultation, a report, no bill

The federal government held a consultation on copyright in the age of generative AI from October 12, 2023 to January 15, 2024. It received about 1,000 responses and 103 submissions from organizations.

The “What We Heard” report, published in February 2025, found limited consensus on the liability of AI companies, but significant interest in transparency requirements. Cultural industries asked that their works be used only with consent, credit and compensation.

Since then, no bill has been tabled. In October 2025, according to a Library of Parliament note, the Minister of AI, Evan Solomon, did not plan to legislate and was instead following the ongoing lawsuits. The federal AI strategy unveiled on June 4, 2026 does not mention copyright, according to an analysis by the law firm McCarthy Tétrault.

Signals, in both directions

The file is not standing still for all that.

• March 2026, Banff summit. The government brought creators and companies together for a first national summit on AI and culture, and created an advisory council on the subject. The Minister of Canadian Identity and Culture, Marc Miller, ruled out loosening the law, saying that the principle of fair compensation for the use of protected content remains unchanged. He also said it would be irresponsible to claim that no action will be taken.
• April 2026, Heritage Committee. MPs adopted a report with 13 recommendations, including creators’ prior consent to model training, transparency about the works used and labelling of AI-generated content.

Rights holders, for their part, speak with one voice. The Coalition for the Diversity of Cultural Expressions calls for authorization, remuneration and transparency. SOCAN asks for “no new copyright exceptions that would permit free, unauthorized use” of works, and its campaign has more than 8,600 signatories. As early as 2025, Copibec regretted the lack of a firm commitment from the government.

The courts are moving faster

In the absence of legislation, judges are drawing the first lines.

In Canada, the Toronto Star, Postmedia, The Globe and Mail, The Canadian Press and CBC/Radio-Canada have been suing OpenAI in Ontario since November 2024. They allege copyright infringement, circumvention of technological protection measures and breach of their websites’ terms of use. On November 7, 2025, the Ontario Superior Court found it had jurisdiction over six OpenAI entities. The trial on the merits is still to come.

Elsewhere, the picture is becoming clearer:

• United States. In June 2025, two federal judges found that training a model on books could qualify as fair use. Judge Alsup, however, denied that protection to the pirated copies downloaded by Anthropic. The company then reached a US$1.5 billion settlement with authors, about $3,000 per work, which received final approval on July 20, 2026.
• European Union. Since August 2, 2025, providers of general-purpose AI models must publish a summary of their training data and adopt a policy to comply with copyright. The Commission has been able to sanction breaches since August 2, 2026.
• United Kingdom. In March 2026, the government dropped its proposed text and data mining exception, with an opt-out for creators, as its preferred way forward.

What it means for rights holders

The American lesson is clear: even where training is found to be lawful, the source of the data matters. Works taken from pirate sites expose those who use them to significant damages. Piracy is therefore not a separate issue from AI: pirate libraries are one of the doors through which works enter the models.

Until there is a Canadian law, three habits protect creators:

• Document where your works circulate, especially on pirate libraries and download sites, with dated evidence.
• Get illegal copies removed: every copy taken down is one less source for unauthorized training.
• Keep a record of your actions: in a lawsuit or a licensing negotiation, that record will have value.

That is exactly what our monitoring does: find copies of your content, get them removed and keep proof of every notice. If you want to know where your works are today, start with our free analysis.

Sources

• Innovation, Science and Economic Development Canada, “What We Heard” report (February 2025): ised-isde.canada.ca
• Library of Parliament, HillNotes (October 27, 2025): hillnotes.ca
• Global News (June 8, 2026): globalnews.ca
• McCarthy Tétrault, on the “AI for All” strategy (June 17, 2026): mccarthy.ca
• Canadian Heritage, AI and culture summit (March 2026): canada.ca
• Noovo Info, statements by Marc Miller (March 17, 2026): noovo.info
• CP24, Heritage Committee report (April 16, 2026): cp24.com
• Coalition for the Diversity of Cultural Expressions (March 18, 2026): cdec-cdce.org
• SOCAN (March 19, 2026): socanmagazine.ca
• Lenczner Slaght, on the jurisdiction ruling in Toronto Star v. OpenAI: litigate.com
• Authors Guild, approval of the Anthropic settlement (July 2026): authorsguild.org
• European Commission, guidelines for general-purpose AI model providers: digital-strategy.ec.europa.eu
• Osborne Clarke, on the UK report (March 25, 2026): osborneclarke.com