In recent years, novel issues arising from the use of AI tools have prompted plaintiffs to file a host of lawsuits against AI companies. Below, Lexpert talks to litigators from Lenczner Slaght LLP, Torys LLP, and Lawson Lundell LLP about the AI litigation landscape in Canada – and the types of claims that could be on the horizon.
No resolution of copyright infringement and terms of service claims to date
In November 2024, the Canadian Legal Information Institute, or CanLII, sued Caseway, a then-newly launched company that described itself as an AI-driven legal research assistant. The nonprofit legal database alleged Caseway had unlawfully bulk downloaded and scraped its content, only to offer it to users who pay a monthly subscription fee. Weeks later, a group of Canadian media companies – including Toronto Star Newspapers Limited, the Canadian Broadcasting Corporation, the Globe and Mail Inc., and Canadian Press Enterprises Inc. – sued OpenAI, alleging the AI giant had been unlawfully scraping their copyrighted content to train its ChatGPT service.
Since then, more lawsuits with similar claims against AI companies have sprung up, including class actions that allege those companies used pirated books or copyrighted artistic works to train generative AI products.
Most of these cases are still pending in court. In January, Caseway founder Alistair Vigier told Canadian Lawyer that the company had reached a settlement with CanLII; the terms of the agreement were not publicly disclosed. Last year, the Ontario Superior Court of Justice issued a decision in the Canadian media case, Toronto Star Newspapers Limited v. OpenAI Inc., affirming that the court has jurisdiction over the case even though OpenAI is based in California; the case is currently before the Ontario Court of Appeal.
However, none of these cases has been resolved on the merits. Sana Halwani, a partner at Lenczner Slaght who represents media companies in the OpenAI case, says she expects to see more copyright infringement cases against AI companies in the near future. In May, she was part of a team of lawyers who filed another copyright infringement lawsuit against OpenAI, this time on behalf of Toronto-based tech company VerticalScope.
Sana Halwani
Both the VerticalScope and media lawsuits claim that OpenAI violated the plaintiffs’ terms of use, which only authorize the use of their content for personal, non-commercial purposes.
“I think we’re still at the leading edge of cases that are going to be brought against large AI companies. I expect you’ll see more of those,” Halwani says. “I also expect that in that context, you will see terms of use violations being alleged in most, if not all, of those cases.
“That’s an interesting area I think the court is going to have to grapple with.”
Andrew Bernstein, a partner at Torys whose litigation practice includes intellectual property disputes, agreed that many copyright claims against AI companies include allegations of terms of service violations. He anticipates seeing more of these disputes – including those involving website terms of service – soon as the use of generative AI tools becomes more common.
“One of the use cases for AI is you take a bunch of documents, and you load it into your AI, and you ask the AI to summarize the documents,” Bernstein says.
Andrew Bernstein
“If you’re downloading 100 news articles, you better look at the terms of service that you have with your news provider and [make] sure that you’re allowed to upload them into your AI, because otherwise, you’re making these copies, and they may not be permitted,” Bernstein says.
“People are just going to have to become a lot more aware than they’ve been in the past of what their rights are… because it’s one thing to pass an article around the office,” he adds. “It’s another to upload it into a third-party piece of software.”
As the tech industry and litigants – including potential plaintiffs – wait for the courts to rule on the merits of these cases, Halwani argues that such a ruling will likely come sooner in non-class actions due to procedural hurdles, such as certification, that can lengthen class actions.
“I think everybody’s watching all of the Canadian cases,” Halwani says. “Anybody who’s involved in these cases is trying to make sure that we’re keeping tabs on what arguments are being made and what, if anything, courts are saying.”
Canadians are filing negligence and product liability claims in US courts
In February, a teenager in Tumbler Ridge, BC, shot and killed her mother and half-brother before returning to her former high school, where she killed six people and injured many more. The shooter was later found dead from a self-inflicted injury.
The following month, Maya Gebala, a 12-year-old who was severely injured after being shot three times, sued OpenAI along with her sister and mother. Filed in the Supreme Court of BC, the lawsuit alleged that approximately 12 OpenAI employees had flagged the shooter’s use of ChatGPT in 2025, which included describing various scenarios involving gun violence to the AI chatbot. The employees recommended informing Canadian law enforcement about the shooter’s ChatGPT activity.
According to the lawsuit, however, OpenAI brushed off the recommendations and instead banned the shooter’s OpenAI account. The shooter then opened a second account, which she used to continue planning a mass shooting and to receive pseudo mental-health counselling and therapy. Gebala and her family sought damages, accusing OpenAI of negligence when it released ChatGPT that prioritized user engagement over safety, prematurely released the chatbot without adequate safety testing, and failed to report the shooter’s activity to law enforcement.
Gebala and her family have since withdrawn their BC lawsuit. In April, Vancouver firm Rice Parsons Leoni & Elliott LLP and US firm Edelson PC announced they were representing the families of multiple Tumbler Ridge victims, including Gebala. The families filed seven separate lawsuits against OpenAI and CEO Sam Altman in the Northern District of California, a US federal court.
Like the BC action, most of the US lawsuits alleged negligence by OpenAI. However, they also claimed the company had violated California’s strict products liability regime, which holds that products are defectively designed when they fail to perform as safely as an ordinary consumer would expect, or when the risks inherent in their design outweigh their benefits. The lawsuits claimed the version of ChatGPT that the Tumbler Ridge shooter had used qualified as defective under both tests.
In April, Edelson PC founder Jay Edelson told the BBC he planned to file more than two dozen legal actions against OpenAI related to the Tumbler Ridge shooting. The two firms noted several reasons for filing the lawsuits in California rather than Canada, including restrictions under BC law that prevented the victims’ estates – and in some cases, the victims’ loved ones – from recovering damages against OpenAI in the province.
Mark Fancourt-Smith, a partner at Lawson Lundell who focuses on commercial disputes, says the US also has a “greater availability of jury trials.” He says he is unaware of any cases in a Canadian court where plaintiffs have brought claims against AI companies, alleging their tools had caused addictive behaviours, self-harm, harm to others, or psychosis.
Last year, an Ontario recruiter who sued OpenAI, claiming ChatGPT induced a weeks-long delusional episode, chose to file his lawsuit in Los Angeles Superior Court. The recruiter’s claims included negligence.
“The cost of damages available to a plaintiff through jury trials in the States is often in excess of what you see awards made in Canada,” Fancourt-Smith says.
Mark Fancourt-Smith
Third-party liability for AI companies?
Over the last few years, the growing availability of generative AI tools – particularly those powered by large language models like ChatGPT or Google Gemini – has changed how many self-represented litigants approach litigation. Fancourt-Smith notes that “the ability to access generative AI, and generate legal pleadings and arguments with the push of a button is, on the one hand… very empowering to self-represented litigants who may not have access to a lawyer.”
On the other hand, however, the lawyer argues that the accessibility of generative AI tools also “facilitates massive filings of suits and generation of incredible amounts of volumes of what look like pleadings that courts have to deal with.”
These issues are highlighted in another US federal court case, Nippon Life Insurance Company of America v. OpenAI Foundation, which was filed in the Northern District of Illinois in March. According to Scott Lucyk, another partner at Lawson Lundell who specializes in commercial disputes and arbitration, the case raises questions about liability and harm to third parties and could serve as a model for similar cases in Canada in the near future.
The case involves Graciela Dela Torre, who had been insured by Nippon Life through her employer. After settling a dispute with her insurer, Dela Torre, who felt misled by her lawyer, uploaded his correspondence to ChatGPT, which validated her distrust. She fired her lawyer, attempted to reopen her case, and relied on legal advice from ChatGPT to file dozens of motions against the insurer, which the courts found served no real legitimate purpose.
Nippon Life then sued OpenAI for tortious interference with a contract, the unlicensed practice of law, and abuse of process. The insurance company argued that ChatGPT unlawfully provided legal advice to Dela Torre without a license, encouraged her to breach her settlement agreement, and helped her abuse the court system. As a result, the insurer was forced to spend significant time and resources litigating claims that had already been settled.
The case is significant because it “involves what I would characterize as alleged harms to third parties,” Lucyk says. “Nippon was a stranger to the initial interaction between ChatGPT [and Dela Torre], but it is now bringing these claims as a party who was impacted by someone’s use of the AI itself.”
Although the case is not in Canada, Lucyk says a ruling could provide insight into how liability works in such a scenario.
“What happens in the United States often follows in Canada, and there’s no reason to think that a similar set of circumstances couldn’t arise in Canada, given the widespread use of these tools, which is only increasing every day,” Lucyk says.