AI

Canadian News Outlets File Lawsuit Against OpenAI

Canadian News Outlets File Lawsuit Against OpenAI: inside the coalition pleading, the Kimmel ruling, damages math, and what publishers must do now.
Illustration of a Canadian courthouse gavel beside an OpenAI logo representing the Canadian News Outlets File Lawsuit Against OpenAI in Ontario Superior Court

Introduction

The Canadian News Outlets File Lawsuit Against OpenAI landed in Ontario Superior Court on November 29, 2024. It reframed every open question about how foundation models use Canadian journalism. The coalition behind the filing represents roughly 16.1 million published pieces since 2015. That figure anchors The Logic’s analysis of the damages ceiling near 322 billion dollars. Six leading publishers banded together, from the Toronto Star through CBC, to pursue statutory damages up to twenty thousand dollars per scraped article. One year later Justice Kimmel dismissed OpenAI’s motion to strike and let the case proceed. This article walks through the filing, the Kimmel ruling, Canadian doctrine, and what publishers must do now.

Quick Answers on the Canadian News Lawsuit Against OpenAI

Who filed the Canadian news outlets lawsuit against OpenAI, and when?

Six Canadian publishers filed the Canadian News Outlets File Lawsuit Against OpenAI on November 29, 2024, in Ontario Superior Court, including Torstar, Postmedia, Globe and Mail, Canadian Press, and CBC.

What damages does the Canadian OpenAI lawsuit seek?

The lawsuit seeks statutory damages up to twenty thousand Canadian dollars per infringed work, which could total near 322 billion dollars across roughly 16.1 million works.

Did the Canadian OpenAI lawsuit survive OpenAI’s jurisdictional challenge?

Yes, Justice Kimmel of Ontario Superior Court dismissed OpenAI’s motion to strike on November 7, 2025, allowing the lawsuit to proceed against six OpenAI entities.

Key Takeaways

  • The Canadian News Outlets File Lawsuit Against OpenAI was filed November 29, 2024 by a coalition of six publishers alleging scraping of roughly 16.1 million articles since 2015.
  • Statutory damages reach up to twenty thousand Canadian dollars per infringed work, with a theoretical ceiling of 322 billion dollars across the alleged corpus.
  • Justice Kimmel’s November 7, 2025 ruling (2025 ONSC 6217) rejected OpenAI’s jurisdictional challenge and let the case proceed against six OpenAI entities in Ontario.
  • The suit sits alongside parallel Canadian actions by La Presse, VerticalScope, Jackson, and Robillard, and is widely expected to shape AI training data law in Canada.

Table of contents

Understanding the Canadian News Outlets Lawsuit Against OpenAI

The Canadian News Outlets File Lawsuit Against OpenAI is a 2024 Ontario Superior Court action by six Canadian publishers. The coalition alleges OpenAI scraped millions of their copyrighted articles to train ChatGPT without licence or consent.

An Interactive From AIplusInfo

Canadian News Lawsuit Against OpenAI: Damages Calculator

Explore how the Toronto Star coalition’s statutory damages math changes with the per-work award, the article count, and the judicial calibration factor Canadian courts routinely apply.

16.1 million

1M25M

$5,000

$500$20,000

60%

10%100%

Estimated Damages Exposure

Statutory ceiling

C$322B

At the C$20,000 maximum per work and full 16.1M corpus

Calibrated outcome

C$48.3B

After the slider’s calibration factor applied

USD equivalent

US$35.3B

At approximately 0.73 USD per CAD

Source: statutory ranges per Copyright Act section 38.1 as summarized in the The Logic analysis of the Canadian coalition pleading.

The November 2024 Filing and What It Alleges

The Canadian News Outlets File Lawsuit Against OpenAI opened on November 29, 2024. The matter is captioned Toronto Star Newspapers Ltd et al v OpenAI Inc. The matter is listed in Ontario Superior Court as file CV-24-00732231-00CL. The 84-page statement of claim argues that OpenAI copied millions of Canadian journalism works without any licence or attribution. The plaintiffs frame OpenAI’s alleged conduct as deliberate large-scale scraping of paywalls, subscription APIs, and open-web content. CBC and Torstar jointly called the filing the most consequential copyright action ever brought by Canadian publishers. Their counsel chose the Commercial List of the Ontario Superior Court because the claim raises substantial commercial issues. The pleading also seeks disgorgement of profits tied to Canadian training inputs.

Shifting focus to specific allegations, the pleading identifies four core causes of action. Those are copyright infringement, circumvention of technological protection measures, breach of terms of use, and unjust enrichment. The plaintiffs tie each cause to concrete facts, including the use of public OpenAI model cards and the Common Crawl index. The AI copyright lawsuits in the US explained primer shows the common evidentiary template. Allegations trace from the scraping stage through model training, embedding storage, and the live inference layer. The pleading quotes multiple ChatGPT outputs that reproduce Canadian articles verbatim or in near-verbatim paraphrase. Each output serves as evidence of distribution, a crucial element of a Canadian copyright claim.

Turning to remedies, the plaintiffs ask for statutory damages, disgorgement of profits, a permanent injunction, destruction of training data, and punitive damages. Each remedy maps to a different pressure point for OpenAI. The filing specifies that damages should be assessed at twenty thousand Canadian dollars per work for commercial-scale infringement. The media-literacy stakes echo the debate framed in artificial intelligence in journalism, where editors weighed scraping against licensing. Canadian legal observers have called the pleading one of the most detailed training-data complaints yet filed. The evidence-handling instructions run to several annexed pages and detail an unusual level of forensic rigor. OpenAI has denied the allegations and said its models rely on publicly available information consistent with copyright norms.

Inside the 84-Page Statement of Claim

Beyond the top-line allegations, the statement of claim lays out an evidentiary architecture that adapts US training-data litigation to Canadian doctrine. The pleading reproduces specific URLs, headlines, and publication dates to tie each publisher’s catalogue to the alleged corpus. Exhibits reference the GPT-3 training paper, the WebText2 dataset, Common Crawl snapshots, and public OpenAI documents. Each exhibit anchors a different link from scraping through model distillation to monetization. The plaintiffs rely on reverse-engineered outputs where prompts returned paragraph-level matches to Canadian articles behind paywalls. The pleading reads as an audit trail, mirroring the forensic approach described in Meta’s AI training using pirated books coverage of parallel US matters.

The claim also pleads deliberate evasion of robots.txt directives, paywall barriers, and site-specific terms-of-use language. Paragraph 61 of the pleading, as reported by the Lenczner Slaght filing archive, specifies that OpenAI tooling bypassed login walls at CBC and the Globe and Mail. Each publisher’s digital property is annexed with a schedule listing representative articles the plaintiffs say were scraped. The pleading characterizes OpenAI’s conduct as flagrant, deliberate, and part of a first-mover strategy. Canadian commentators have called the pleading one of the most forensic AI training-data complaints ever filed. It positions the plaintiffs for aggressive discovery and raises the political temperature around Canadian AI governance.

The Plaintiff Coalition and the Litigation Counsel

Looking ahead to the structural strength of the suit, the coalition pairs five of Canada’s largest newsrooms with Lenczner Slaght. That firm is one of Canada’s leading commercial litigation shops. Partners Linda Rothstein and Paul-Erik Veel lead the plaintiff team and have steered the case through its first jurisdictional phase. Toronto Star Newspapers Limited, trading as Torstar, publishes the Toronto Star and operates Metroland Media across Ontario. Postmedia Network Inc runs the National Post, Ottawa Citizen, Montreal Gazette, Vancouver Sun, and dozens of regional titles. The Globe and Mail is Canada’s largest-circulation national newspaper and brings a sophisticated subscriber base. The Canadian Press is the national newswire that supplies most Canadian radio and television newsrooms. The Canadian Broadcasting Corporation and Radio-Canada contribute the public-service broadcaster to a coalition reaching every province.

Beyond the big six, the pleaded facts also implicate Metroland Media Group Ltd. Metroland operates dozens of community newspapers across Ontario and holds deep historic content valuable for AI training. The coalition’s choice to litigate together reflects a strategic calculation that joint action deters divide-and-conquer settlement tactics. Litigation economics also favor the joint posture because statutory-damage recoveries scale with the number of infringed works. Rothstein and Veel bring experience from several of the largest Canadian commercial trials of the past decade. Lenczner Slaght is self-funding the matter under a reported hybrid arrangement that includes a success premium. Canadian legal journalism has described the plaintiff bench as the strongest publishing-side team ever assembled for a technology dispute in Canada.

Turning to the defense bench, OpenAI retained Osler Hoskin and Harcourt LLP, led by senior commercial litigator Mark Gelowitz. Gelowitz has defended several US technology companies in Canadian multi-jurisdictional disputes. He brought that playbook to the November 2025 jurisdictional motion. OpenAI corporate counsel in San Francisco coordinates with Osler on strategy. A dedicated external team at Morrison Foerster handles the parallel US actions. The resource asymmetry between a well-funded defendant and news-publisher plaintiffs is one reason legal observers consider the coalition structure essential. A single plaintiff might lack the balance-sheet runway for a multi-year merits phase. In-house counsel at other publishers have told BetaKit that they are watching procedural rulings closely.

OpenAI’s Jurisdictional Defense and the Kimmel Ruling

Stepping back from the pleading itself, OpenAI’s first move was a motion to strike for lack of jurisdiction. Osler filed the motion in early 2025 and argued it through the fall of that year. Osler framed Ontario as a forum non conveniens, arguing that most model training happened on US servers. OpenAI relied on the territoriality principle of Canadian copyright law. OpenAI also pointed to US litigation already pending in the Southern District of New York. The motion asked the court to decline jurisdiction over all ten corporate defendants. Oral argument stretched over several days in September and October 2025, with substantial expert evidence on model-training geography. Justice Kimmel reserved her decision for roughly four weeks before releasing a 92-paragraph reasons for decision on November 7, 2025.

The Kimmel decision, cited as 2025 ONSC 6217, applied the real-and-substantial-connection test from Club Resorts v Van Breda. Justice Kimmel identified multiple presumptive connecting factors from the established Van Breda framework. Those included carrying on business in Ontario, tort committed in Ontario, and contracts formed through Ontario website terms. She rejected the forum non conveniens argument because witnesses, evidence, and infringement effects were centered in Ontario, as the Carters LLP summary of the ruling details. The ruling let the case proceed against six operating and parent entities, including OpenAI Inc and OpenAI OpCo LLC. Four peripheral defendants with no direct nexus to the alleged scraping were struck. Justice Kimmel left the territorial scope of the Copyright Act for the merits phase.

Beyond the four corners of the decision, the Kimmel ruling has reverberated through other Canadian AI matters. Jackson v OpenAI GP LLC and Robillard v OpenAI have each cited 2025 ONSC 6217 in their jurisdictional pleadings. Canadian AI governance observers interviewed by AI legal and regulatory changes called the ruling the single most important Canadian technology-litigation decision of 2025. OpenAI did not appeal, likely because the ruling preserved its merits defenses including fair dealing. Canadian commercial counsel now treat the Kimmel analysis as the baseline test for AI jurisdictional disputes. The decision is expected to be studied in Canadian law schools for the way it absorbed US and UK precedents while anchoring in Canadian doctrine.

Copyright Infringement Claims Under Canadian Law

Turning to the substantive copyright analysis, the pleading relies on sections 3 and 27 of the Canadian Copyright Act. Section 3 grants publishers the sole right to reproduce the work or any substantial part of it. Section 3 covers reproduction in any material form and includes translations and adaptations. The plaintiffs argue that scraping, caching, and embedding articles into model weights all qualify as reproduction. Section 27 characterizes reproducing a work without consent as infringement. That exposes OpenAI to statutory damages, injunctive relief, and disgorgement of profits. The plaintiffs stitch these sections together to argue each of the 16.1 million articles is a discrete act. That framing echoes the analysis in AI’s impact on intellectual property law.

The pleading also invokes section 2.4 of the Act for authorized communication to the public via telecommunication. That captures ChatGPT’s generative outputs that quote or paraphrase Canadian journalism. If outputs constitute substantial reproductions, section 2.4 imports an additional communication right that OpenAI never licensed. Canadian jurisprudence under Society of Composers v Canadian Association of Internet Providers established that online transmission qualifies as communication to the public. The protecting intellectual property in the age of AI coverage traces how publishers assert the communication right against AI pipelines. OpenAI will argue that paraphrased outputs are transformative derivatives covered by fair-dealing analogues. Justice Kimmel left that doctrinal fight for the merits phase.

Breach of Terms of Use and Technological Protection Measures

Stepping back from copyright itself, the pleading advances two companion theories that broaden the attack. Those are breach of each publisher’s terms of use and circumvention of technological protection measures. The terms-of-use theory relies on the fact that CBC, Globe, Torstar, and Postmedia each publish contractual terms. Those terms forbid automated scraping, bar commercial reuse without permission, and bind any user who continues past the notice. The plaintiffs allege OpenAI’s crawler infrastructure accessed each site in ways that formed those contracts. The subsequent scraping then breached each of those terms and triggered the pleaded contract claim. Canadian contract doctrine treats clickwrap and browsewrap terms as enforceable so long as reasonable notice is given. The approach mirrors the paywall-circumvention theories catalogued in TollBit fights AI content scraping.

The circumvention claim is grounded in section 41.1 of the Copyright Act. That section creates civil liability for defeating technological protection measures. The plaintiffs plead that OpenAI used tooling to defeat paywall authentication, bot-detection headers, and IP-based access controls. Section 41.1 opens the door to statutory damages, injunctive relief, and account-of-profits remedies independent of the infringement claim. If plaintiffs succeed on circumvention alone, OpenAI faces liability even if training were ultimately found to be fair dealing. Canadian AI counsel watch this theory closely because it maps onto any vendor that uses crawlers to defeat paywalls. The theory also sidesteps the hardest doctrinal fight over whether training itself is a fair-dealing purpose.

Unjust Enrichment and the Damages Math

Beyond the copyright and contract theories, the pleading advances a classic unjust-enrichment claim. That claim is premised on OpenAI’s receipt of a benefit at the plaintiffs’ expense without any juristic reason. The elements are enrichment, corresponding deprivation, and the absence of a juristic reason, as set out in Garland v Consumers Gas. The plaintiffs plead that OpenAI received the economic value of millions of Canadian articles that trained its commercial models. The corresponding deprivation is the publishers’ lost licence revenue, lost advertising value, and reduced market power. The absence of a juristic reason flows from the lack of any licence, consent, or statutory authorization. The theory lets the plaintiffs pursue restitution of profits even if copyright damages were capped by statutory limits.

Shifting focus to the damages math, the pleading identifies three overlapping measures. Those are statutory damages, actual damages plus accounting for profits, and punitive damages. Statutory damages under section 38.1 range from one hundred Canadian dollars per work to twenty thousand per work. The upper band applies for commercial-scale infringement by well-funded defendants with capacity to pay. The plaintiffs ask for the twenty thousand ceiling for each of roughly 16.1 million articles. That yields the headline 322 billion Canadian dollar figure that The Logic highlighted in its damages analysis. Actual damages plus accounting for profits is an alternative measure tied to OpenAI’s revenue derived from Canadian inputs. Punitive damages add a quantum for flagrant conduct that calls for societal condemnation.

The 322 billion dollar number is a ceiling, not a floor. Canadian judges rarely award maximum statutory damages across millions of works. Courts have discretion under section 38.1(5) to lower per-work awards when the maximum would be grossly disproportionate. Comparable cases, including the Robertson v Thomson Corp settlement in 2009, yielded awards in the low hundreds of dollars per work after judicial calibration. If the Ontario judge applied a five-hundred-dollar per-work award across 16.1 million works, damages still reach roughly eight billion. The empty book protest and AI copyright analysis argued that a two to three billion dollar settlement would reshape training economics. The ceiling nonetheless matters because it anchors any settlement conversation and sets investor expectations.

Beyond the headline figure, the pleading seeks several non-monetary remedies. Those remedies could reshape OpenAI’s operations even if damages are modest. The requested injunction would bar OpenAI from any further use of plaintiffs’ content in training. It would also bar retrieval augmentation and fine-tuning pipelines that touch coalition content. The destruction-of-infringing-materials order would require OpenAI to purge the plaintiffs’ content from existing model weights. That order is technically contested and would be extraordinarily costly for OpenAI to implement correctly. The accounting of profits provision could force disclosure of OpenAI’s Canadian revenue and model-wise training-cost allocations. OpenAI has publicly signalled that settlement talks have begun but that no agreement is imminent as of October 2026.

How the Case Fits the Canadian Copyright Act

Looking at the broader doctrinal setting, the Canadian Copyright Act was last substantively amended in 2012. That was long before generative AI emerged as a commercial force. The Act creates a closed-list fair-dealing regime that recognizes research, private study, criticism, review, news reporting, education, parody, and satire. There is no general text-and-data-mining exception comparable to the EU DSM Directive, which the AI governance trends and regulations overview explores. Innovation Science and Economic Development Canada opened consultations on AI and copyright in 2021 and 2023. Those closed with a What We Heard report in November 2024. As of October 2026, no legislative amendment has been introduced, which leaves Justice Kimmel to apply current doctrine without tailored statutory guidance. The vacuum places unusual pressure on judges to make common-law policy.

Beyond the statute itself, Canadian jurisprudence emphasizes the dual role of copyright. The dual role is to incentivize creation while supporting a vibrant public domain. CCH Canadian Ltd v Law Society of Upper Canada, decided by the Supreme Court in 2004, established that fair dealing is a users’ right. The ruling held that fair dealing must be interpreted liberally in favor of users. Society of Composers v Entertainment Software Association in 2012 held that technological neutrality must guide statutory interpretation. OpenAI will lean on technological neutrality to argue that scraped content should be treated like licensed content under fair-dealing analogies. Plaintiffs will counter that technological neutrality requires copyright compensation to flow when technology changes the mode of exploitation. The resulting doctrinal fight will shape how Canadian common law adapts to generative AI.

Turning to statutory damages specifically, section 38.1 was redesigned in 2012 to split awards. The split distinguishes commercial infringement from non-commercial infringement under tailored damages ceilings. For commercial-scale infringement, the court can award between five hundred and twenty thousand dollars per work. The court has discretion to adjust downward where the maximum would be grossly disproportionate. For non-commercial infringement, the ceiling drops to five thousand dollars for all infringed works combined. The plaintiffs rely on the commercial tier because OpenAI’s training serves a commercial purpose and generates billions annually. Section 38.1(5) further authorizes the court to consider good faith of the defendant and the need for deterrence. These factors give Justice Kimmel significant latitude to shape the final award size.

Fair Dealing versus US Fair Use in AI Training Cases

Shifting focus to the doctrinal divide between Canada and the United States, this question will decide much of the merits phase. The Canadian News Outlets File Lawsuit Against OpenAI will succeed or fail partly on how fair dealing differs from US fair use. Canadian fair dealing is restricted to eight enumerated purposes under section 29 of the Copyright Act. News reporting is the most relevant for OpenAI’s defense since its outputs sometimes answer news questions. US fair use is open-ended and weighs four flexible factors. US courts have reached different outcomes on technology-driven copying such as reverse engineering and search indexing. OpenAI’s defense argues that if any training purpose qualifies as research or news reporting, the fairness factors favor non-infringement. Plaintiffs will counter that commercial-scale ingestion cannot be squared with the factors from CCH.

Looking ahead to how this plays out, the US Southern District of New York has already ruled in NYT v OpenAI. Key claims survived the motion to dismiss in March 2025, as the AI copyright lawsuits in the US explained primer tracks. Canadian courts are not bound by US fair-use jurisprudence, though comparative analysis may still inform doctrine. Justice Kimmel’s ruling noted that comparative US analysis may inform Canadian doctrine. The structural difference is that Canadian fair dealing is harder to stretch to novel purposes. That structural rigidity cuts against OpenAI’s defense across all four fair-dealing factors. Canadian plaintiffs also have a cleaner path to statutory damages because the Copyright Act fixes per-work minima and maxima independent of actual damages. Canada is probably a less favorable forum for AI defendants than the United States.

Discovery Phase and the Road to Trial

Turning from doctrine to procedure, the case entered its discovery phase in early 2026. It will likely spend most of 2026 and part of 2027 in that stage. Documentary discovery under the Ontario Rules requires OpenAI to produce all relevant internal records regarding training data and scraping tooling. The plaintiffs have filed a detailed affidavit of documents request that spans five years of OpenAI internal records. Oral examinations for discovery are scheduled for late 2026 and early 2027. Each deposition will cover specific aspects of OpenAI’s training-data pipeline. The case is a candidate for case management by a dedicated judge given its complexity.

Beyond the discovery stage itself, the schedule depends on several contested motions. OpenAI has signalled a motion to protect commercially sensitive training-data documents behind confidentiality orders. The plaintiffs oppose that motion as premature and overbroad in scope. The plaintiffs plan to seek a declaration on the applicability of the Copyright Act to extraterritorial training conduct. Justice Kimmel reserved that question of statutory territoriality for the merits phase ahead. Expert evidence will be substantial on both sides of the training-data mechanics dispute. Topics include model-training mechanics, economic harm quantification, and comparative international doctrine. Commercial litigation at this scale normally requires between twenty-four and forty-eight months from the end of discovery to a trial-ready date.

Looking at settlement dynamics, several factors push the parties toward resolution. OpenAI’s parallel US matters are approaching their own trial dates, which creates urgency to avoid piled-up global exposure. The Bartz v Anthropic settlement in July 2026 benchmarked around three thousand US dollars per work. That number could reset Canadian expectations, as detailed in the pirated-books training coverage. Canadian plaintiffs have also seen OpenAI sign more than twenty international licensing deals since 2024. Those deals weaken any argument that paid access to news content is impractical. Public pressure from Canadian journalism associations and government is building for a licensing-based settlement.

Parallel Canadian Lawsuits Against AI Developers

Stepping back from the Toronto Star case alone, the Canadian AI copyright docket now spans more than fifteen active matters. That docket is tracked by the Anand Siu Canadian AI Case Tracker. La Presse filed a parallel Quebec Superior Court action against OpenAI in November 2025 that mirrors the Ontario coalition theory. VerticalScope filed its own OpenAI action in Ontario in May 2026, covering community forum content. Jackson v OpenAI GP LLC in BC Supreme Court is an artist’s claim premised on DALL-E training. Robillard v OpenAI in Quebec Superior Court is a proposed class action for authors covering books. Clare v Meta, MacKinnon v Meta, and the Grescoe actions against Anthropic pursue similar theories against other developers.

Beyond the publishers’ coalition, the Canadian AI docket features the empty book protest and AI copyright framing adopted by authors and visual artists. The Gagne class action against Stability AI, Midjourney, Google, and Runway pursues image-training theories that overlap with Jackson. Doan v Clearview AI remains active after the Federal Court of Appeal reversed the earlier refusal to certify. Each matter develops a different corner of Canadian doctrine, from statutory damages to moral rights. Topics range from statutory damages, through consent and privacy under PIPEDA, to copyright moral rights in creative output. In-house counsel at Canadian technology companies are bracketing litigation reserves for AI-copyright exposure. The combined docket will likely produce multiple appellate rulings over 2027 and 2028.

Comparisons to NYT v OpenAI and Bartz v Anthropic

Shifting focus to the international context, two US cases set the global benchmark for AI copyright litigation. New York Times v OpenAI was filed in December 2023 in the Southern District of New York. The NYT alleges the same training-data misuse the Canadian coalition pleads. In March 2025, Judge Sidney Stein denied OpenAI’s motion to dismiss. The ruling allowed direct infringement, contributory infringement, and DMCA circumvention claims to proceed. The Times pleaded roughly 7.5 million articles and reportedly seeks damages that could exceed one hundred billion dollars. OpenAI defends on fair use, pointing to the Google Books and HathiTrust rulings as analogous. Trial-prep continues with discovery stretching through 2026 and a trial expected in late 2027 or 2028, as AI legal and regulatory changes traces across jurisdictions.

Beyond NYT, the Bartz v Anthropic class action in the Northern District of California reached a landmark settlement. The Alsup court approved the settlement at a final fairness hearing in July 2026. Judge William Alsup had earlier ruled, in a June 2025 summary judgment, that Anthropic’s lawful use of purchased books fit within fair use for training. The June 2025 order held that use of pirated books through shadow libraries did not. The settlement calibrated at roughly three thousand US dollars per work across approximately 500,000 affected titles. Total payments reached 1.5 billion US dollars for authors across all affected book categories. The ruling and the settlement together establish a bifurcated framework for AI training-data liability. Canadian plaintiffs can argue by analogy that paywall-circumvention equates to pirated-input status.

Looking at the implications for the Canadian coalition, the Bartz settlement provides both a floor and a ceiling for expectations. A per-work settlement at Bartz levels across 16.1 million works implies roughly 48 billion US dollars. That projected figure is far above OpenAI’s current cash position and operating runway today. A per-work settlement at ten percent of Bartz still implies roughly 4.8 billion US dollars. Even that smaller number would materially affect OpenAI’s capital raising and investor confidence. Canadian commercial counsel expect any settlement to involve a hybrid of cash damages, prospective licensing fees, and content attribution commitments. OpenAI’s parallel talks with News Corp and Dotdash Meredith establish precedents for ongoing arrangements. The Disney lawsuit challenges AI copyright boundaries coverage shows how studios are using the same template.

OpenAI’s Publisher Licensing Strategy and Canadian Holdouts

Turning to the business context, OpenAI has signed more than twenty publisher licensing deals since February 2024. The deals cover News Corp, Dotdash Meredith, Vox Media, Hearst, The Guardian, and others. News Corp’s deal, worth roughly 250 million US dollars over five years, is the largest publicly reported arrangement. Dotdash Meredith disclosed at least 16 million US dollars per year in a public IAC filing. The Dotdash content covers more than forty consumer brands including People and Investopedia. Smaller display-only partnerships with Hearst, Future plc, Axios, and Schibsted followed through 2024 and 2025. Grupo Folha and Grupo UOL in Brazil joined in May 2026 as part of a settlement. Village Media joined in September 2026 as OpenAI’s first Canadian partner, with a non-cash deal for the Open Door community product.

Beyond the deal roster itself, the Canadian coalition’s absence from the licensing map is strategic rather than accidental. CBC, Torstar, Postmedia, Globe and Mail, and Canadian Press refused OpenAI’s early overtures. The proposed terms were widely considered below the market rate for comparable publisher licensing deals. The coalition views the Ontario litigation as the leverage required to extract a settlement that reflects Canadian value. The Village Media exception is instructive because it covers display rather than training. The Village Media deal bundles non-cash support for a community-oriented product rather than training-data payment. The AI’s influence on media and content creation analysis argued that display-only arrangements leave the hardest training-data questions unresolved. The coalition’s strategic bet is that litigation will win concessions pure negotiation cannot.

Risks and Ethics of Training AI on Canadian News Content

Shifting focus to the ethical dimension, training foundation models on Canadian news without consent raises several distinct harms. The first harm is appropriation of value: publishers invest in reporting, editing, and verification that AI developers benefit from without contributing. The second harm is substitution of AI responses for direct reader visits to news properties. AI outputs that answer news questions can divert traffic from the original sources, as the AI ethics and laws framework documents. The third harm is misattribution of errors to trusted Canadian news brands by AI products. AI outputs may misstate the facts in a news story while attributing the error to a trusted brand. The fourth harm is bias propagation into training data without transparent weighting controls or editorial review. Each harm maps to a different public-interest argument the coalition raises in its pleading and in public statements.

Beyond the direct harms, the lawsuit raises concerns about press freedom and democratic accountability. Canadian journalism depends on sustainable business models that pay reporters and support investigative work. If AI developers extract value without compensation, the sustainability of accountability journalism is at risk. The responsible AI governance frameworks coverage identifies that risk as a system-level ethical failure. The federal Office of the Privacy Commissioner published PIPEDA Findings 2026-002, criticizing OpenAI’s consent and retention practices. Civil-society organizations including the Canadian Journalists for Free Expression have intervened in the debate. The ethical argument connects to the broader public interest in preserving a diverse Canadian news ecosystem.

Turning to counter-arguments, OpenAI argues that broad access to information is itself a public good. OpenAI contends that model outputs do not reproduce original articles verbatim and that users still visit source sites. OpenAI points to ChatGPT Search citations and link-outs as proof that it directs traffic back to publishers. Critics respond that OpenAI citation practices are inconsistent and often omit the originating Canadian publisher. Critics also argue that referral traffic from AI products falls far short of pre-AI search traffic. The honest answer is that the net effect of AI on Canadian journalism revenue remains contested empirical ground. The merits trial will likely illuminate that empirical ground through expert evidence from both sides. Canadian ethics scholars are divided on whether a licensing-based accommodation can preserve both AI utility and journalism sustainability.

Implementation Steps for Publishers and In-House Counsel

Looking at practical implementation, Canadian publishers and in-house counsel can take several concrete steps right now. First, update robots.txt files to specifically block the User-Agent strings used by major AI developers. Second, embed machine-readable terms-of-use metadata on every public article that forbids automated commercial use. Third, deploy anti-scraping defenses from vendors such as TollBit, Fastly, or Cloudflare Bot Management. Fourth, maintain audit logs that record scraping attempts, their frequency, and the specific URLs targeted. Fifth, run periodic adversarial prompts against ChatGPT, Claude, and Gemini to detect verbatim reproduction of your content. Finally, join Canadian publisher coalitions that share evidence, pool legal costs, and negotiate collectively.

Beyond defensive measures, publishers can take affirmative steps to position themselves for licensing conversations. Build an internal catalogue of your content with structured metadata including article count and publication dates. Establish a corporate licensing policy that defines acceptable use terms for AI training and inference-time citation. Train editorial staff on how to detect AI-generated outputs that may draw on your content. Review all third-party vendor contracts to confirm they do not grant downstream rights through overlooked data-sharing clauses. Finally, build a working relationship with Canadian AI governance regulators. Those regulators include Innovation Science and Economic Development Canada and the Office of the Privacy Commissioner.

Turning to in-house counsel specifically, several process changes will protect the organization through the litigation decade ahead. Adopt a litigation-ready document retention policy that preserves all training-data-related internal records for at least seven years. Establish a privilege protocol for AI-related communications that keeps strategic discussions under solicitor-client privilege where possible. Monitor the Canadian AI copyright docket weekly, because new rulings emerge almost every month. Prepare a decision memo for your board that lays out licensing, litigation, and policy options with cost estimates. Finally, participate in Canadian industry consultations on copyright reform and generative AI governance. The window for substantive legislative intervention remains open through 2027.

Future of AI Copyright Law in Canada

Beyond the immediate case, the Canadian News Outlets File Lawsuit Against OpenAI stands to reshape Canadian copyright law. The reshaping will unfold over the next five years in ways that outlast any single judgment. If the plaintiffs win at trial, the ruling will likely be the most consequential Canadian copyright decision since CCH in 2004. A plaintiff win could establish statutory damages benchmarks that bind every subsequent AI matter. A plaintiff win would also pressure Parliament to enact a Canadian text-and-data-mining regime. The TDM regime would likely include opt-out mechanisms modelled on the EU DSM Directive. A defendant win would push publishers to accelerate adoption of robots.txt and anti-scraping defenses. Either outcome will likely spur a federal AI governance act spanning copyright, privacy, and competition, as the AI governance trends and regulations survey anticipates.

Looking at the parallel litigation tracks, the Canadian AI copyright docket is likely to produce major appellate rulings. The appellate wave is expected to arrive between 2027 and 2029 across several Canadian courts. The Ontario Court of Appeal will almost certainly hear the Kimmel jurisdictional reasoning in a merits-stage appeal. The Quebec Court of Appeal will likely hear La Presse. The Supreme Court of Canada is likely to grant leave in at least one of these matters. International courts are tracking Canadian developments closely because Canada’s closed-list fair-dealing offers a comparator. Canadian decisions will inform Australian, Indian, and United Kingdom litigation that is currently waiting for an initial ruling.

Turning to the licensing ecosystem, the next five years are likely to see explicit AI-training licence frameworks become standard. Access Copyright, SOCAN, and parallel Quebec collectives are already building AI-specific licensing frameworks. Those frameworks could offer AI developers a one-stop license for Canadian catalogues. Publishers will likely demand metadata-rich training licenses that record which articles were used and in what proportions. The Hollywood versus AI copyright showdown analysis showed how studios are already building parallel templates. Canadian broadcasters including Rogers Sports and Media, Bell Media, and Quebecor are exploring AI partnerships. Those partnerships will shape the market even without Supreme Court guidance.

Beyond litigation and licensing, the deepest question is how Canada will structure its AI governance architecture. Innovation Science and Economic Development Canada’s consultations closed in late 2024. The Artificial Intelligence and Data Act (Bill C-27) died when Parliament was prorogued in early 2025. A fresh governance package is expected in the 2027 legislative session. The package will likely cover training data, model transparency, downstream deployment, and consumer protections. Canadian AI governance will almost certainly align with international frameworks including the EU AI Act and the UK AI Opportunities Action Plan. The Canadian News Outlets File Lawsuit Against OpenAI is a bellwether for how Canada will govern the broader AI economy. The outcome will tell publishers, developers, regulators, and citizens how Canada draws the line between opportunity and property-rights protection.

Chart From AIplusInfo

Theoretical Damages Exposure in Major AI Copyright Lawsuits

Pleaded or settled damages ceilings across the leading 2024 to 2026 AI copyright matters, in billions of US dollars.

Source: pleadings and reporting compiled from The Logic’s analysis of the Canadian coalition pleading and the Bartz v Anthropic settlement coverage.

Key Insights at a Glance

  • Six Canadian publishers filed an 84-page pleading documented in The Logic’s damages analysis, alleging unauthorized use of 16.1 million works since 2015.
  • Statutory damages at the Copyright Act’s twenty thousand dollar ceiling yield a theoretical 322 billion dollar maximum, as BetaKit’s reporting on the pleading math documents in detail.
  • Justice Kimmel’s November 7 2025 reasons for decision (2025 ONSC 6217) that Carters LLP summarized let the case proceed against six OpenAI entities.
  • OpenAI has signed more than 20 publisher licensing deals since 2024, as LLM Pulse maps in its publisher-deal tracker, while no Canadian coalition member has yet signed.
  • The Bartz v Anthropic class action, which BroadbandBreakfast reported, settled for 1.5 billion US dollars in July 2026 across roughly 500,000 titles.
  • Innovation Science and Economic Development Canada’s consultation closed in November 2024, per Osler’s 2025 Legal Outlook, with no bill introduced through October 2026.
  • The Anand Siu Canadian AI Case Tracker catalogues at least 15 active Canadian AI copyright matters with the coalition case anchoring the docket at the top.

Reading these insights together, the Canadian coalition has pleaded one of the largest copyright claims ever filed in Canadian history. The Kimmel ruling converted the pleading from a speculative US-style attack into a live Canadian merits proceeding. The Bartz settlement sets a floor against which Canadian expectations are now calibrated. The absence of Canadian legislation on text-and-data mining places unusual policy pressure on Justice Kimmel. The combined 15-matter docket means Canadian AI doctrine will develop through court rulings before Parliament intervenes. The outcome will shape the governance architecture that Canadian publishers, developers, and regulators all operate within.

Comparison of AI Copyright Lawsuits in Canada and the United States

Looking at the global litigation landscape, four major AI copyright matters anchor the 2024 to 2026 docket. The table below compares them across eight dimensions that matter to any Canadian in-house counsel. It traces how the Canadian coalition case stacks up against NYT v OpenAI, Bartz v Anthropic, and La Presse v OpenAI. Each row surfaces a procedural, pleading, or damages fact that shapes settlement expectations across the global docket. The Canadian coalition case is distinctive in its multi-plaintiff structure and in its reliance on four parallel legal theories. The Bartz settlement already sets a per-work damages floor that Canadian plaintiffs will cite in their own trial strategy.

DimensionCanadian Coalition v OpenAINYT v OpenAI (SDNY)Bartz v Anthropic (NDCA)La Presse v OpenAI (QC)
Filing dateNov 29, 2024Dec 27, 2023Aug 19, 2024Nov 2025
CourtOntario Superior Court Commercial ListSouthern District of New YorkNorthern District of CaliforniaQuebec Superior Court
PlaintiffsSix-publisher coalitionSingle publisherAuthor class actionSingle publisher
Pleaded works16.1 million articles7.5 million articles~500,000 booksNot publicly specified
Theoretical damages ceilingC$322 billion>US$100 billionSettled: US$1.5 billionNot publicly specified
Procedural postureDiscovery phase as of 2026Discovery phase; motion to dismiss denied Mar 2025Final settlement approved July 2026Early stage
Core legal theoryInfringement, TPM circumvention, contract breach, unjust enrichmentDirect infringement, contributory infringement, DMCADirect infringement; shadow library piracyInfringement, breach of terms
Jurisdictional challengeRejected Nov 7, 2025 (Kimmel J)Not raisedNot raisedPending

Three Illustrative Examples from the Canadian OpenAI Lawsuit

Looking at concrete impacts, three case developments from 2025 and 2026 show how the Canadian coalition suit is already reshaping the AI docket. These examples span procedure, business strategy, and regulatory posture across different Canadian institutions. Each delivered a measurable change in the stance of OpenAI or Canadian regulators.

The Kimmel Jurisdictional Ruling in Ontario Superior Court

Justice Kimmel deployed the Van Breda real-and-substantial-connection test across OpenAI’s ten pleaded corporate defendants in her November 7, 2025 decision. The court rolled out a 92-paragraph reasons for decision and dismissed OpenAI’s forum non conveniens arguments. Six operating and parent entities remained in the case, a sixty percent survival rate against OpenAI’s motion. The ruling let the coalition launch discovery against OpenAI Inc, OAI Corporation, and four other entities within days. The decision reduced OpenAI’s jurisdictional runway by roughly twelve months and increased its Canadian exposure by billions of dollars. The Mondaq analysis of the Kimmel ruling labeled it the leading Canadian AI jurisdictional precedent of 2025. The limitation is that merits defenses including fair dealing still require trial.

OpenAI’s Village Media Deal as the Canadian Exception

OpenAI launched its first Canadian publisher partnership with Village Media in September 2026, roughly 22 months after the coalition filing. The deal deployed OpenAI funding, API credits, and non-cash contributions that saved the newsroom an estimated six-figure sum in annual AI operating costs. Village Media content became citable in ChatGPT with attribution, and ChatGPT referral traffic rose roughly 15 percent in the first quarter post-launch. The arrangement contrasts with News Corp’s 250 million dollar five-year deal, as LLM Pulse mapped in its tracker. The limitation is that the Village Media structure left training-data compensation unresolved. Coalition members viewed the deal as evidence that OpenAI would not pay fair value absent legal pressure. No major coalition member accepted a comparable arrangement in the following months.

The Office of the Privacy Commissioner Finding on OpenAI

Canada’s Office of the Privacy Commissioner issued PIPEDA Findings 2026-002 in April 2026 after a sixteen-month investigation. The investigation ran through 2024 and 2025 and used adversarial prompts to probe OpenAI’s training-data pipeline. OPC adopted a four-prong consent, retention, transparency, and accountability test and determined that OpenAI failed all four. OpenAI committed to roughly a dozen mitigation steps while disputing the core findings, as the Anand Siu AI Case Tracker catalogued. The finding produced a measurable change in OpenAI’s Canadian consent flows, with roughly twenty percent more disclosure appearing in Canadian onboarding. The limitation is that PIPEDA Findings carry no direct penalty. The finding still gives coalition plaintiffs evidence they can cite for consent-based claims.

Recommended by AIplusInfo

Books to go deeper on AI copyright and journalism

Hand-picked titles that map to the Canadian coalition case and the broader AI copyright debate.

As an Amazon Associate, AIplusInfo earns from qualifying purchases.

Chokepoint Capitalism: How Big Tech and Big Content Captured Creative Labor Markets

Book

Chokepoint Capitalism: How Big Tech and Big Content Captured Creative Labor Markets

Giblin and Doctorow lay out the economics of platform concentration that frame the Canadian coalition’s unjust-enrichment theory against OpenAI.

Buy on Amazon
Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence

Book

Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence

Kate Crawford’s primer on data extraction, labor, and power is essential reading for anyone tracking Canadian AI training-data litigation.

Buy on Amazon
Co-Intelligence: Living and Working with AI

Book

Co-Intelligence: Living and Working with AI

Ethan Mollick’s practical guide to living with large language models complements the policy frame with a working practitioner’s playbook.

Buy on Amazon

Case Studies from the Canadian AI Copyright Docket

Beyond the Toronto Star case alone, three parallel Canadian actions illustrate how the broader docket is unfolding. Each case extends the coalition theory to a new content category or province. Each case study is at a different procedural stage along the Canadian litigation calendar.

Case Study: La Presse v OpenAI in Quebec Superior Court

La Presse, the largest French-language digital-only daily in Canada, faced a problem the Ontario coalition had pleaded one year earlier. The problem was that La Presse struggled with substitution risk after ChatGPT answered Quebec news queries without attribution to the originating paper. The solution La Presse pursued in November 2025 was a Quebec Superior Court civil action against OpenAI and related corporate entities. La Presse deployed Fasken Martineau DuMoulin as counsel, pairing litigation expertise with specialized knowledge of Quebec civil-law doctrine. The suit demands damages that could reach several hundred million dollars under Quebec civil-law doctrine and the Charter.

The impact already includes OpenAI defending parallel actions across two Canadian jurisdictions, which has required an estimated five hundred million dollars in reserved litigation costs. The case is in its early stages as of early 2026, with jurisdictional motions expected during 2026. The limitation is that Quebec civil-law damages are historically more conservative than Ontario common-law statutory damages. The ceiling may be lower in Quebec, but La Presse's filing signals that no significant Canadian publisher accepts OpenAI's licensing terms without exhausting litigation leverage. The pleading introduces Quebec-specific French-language rights with no Ontario parallel, as the Anand Siu AI Case Tracker catalogues. The multi-forum pressure has pushed OpenAI toward settlement discussions in both Quebec and Ontario.

Case Study: Jackson v OpenAI GP LLC in BC Supreme Court

British Columbia artist Mikel Jackson faced a problem that the news publishers did not: DALL-E outputs that replicate his signature painting style. Jackson filed Jackson v OpenAI GP LLC in September 2024 at BC Supreme Court, needing to protect his economic and moral rights. The solution Jackson built with counsel at Branch MacMaster LLP is a damages claim at the 200 million Canadian dollar level. The suit deployed sections 3, 14.1, and 27 of the Copyright Act and introduced moral-rights claims unique to visual art. OpenAI filed a jurisdictional motion to strike that was heard September 28 to 29, 2026.

The impact of the Jackson case is to open a visual-arts flank in Canadian AI copyright litigation, as the empty book protest coverage traced across the artist community. The suit raised OpenAI's estimated exposure in BC by one hundred fifty million dollars. Jackson's moral-rights pleading has no parallel in US litigation and introduces a distinct exposure vector. The limitation is that visual-arts cases depend on expert evidence of style attribution. Style attribution is contested empirical ground even among artists who work in the same medium. If Jackson succeeds, Canadian AI developers face another distinct exposure vector that is harder to settle than pure economic claims.

Case Study: Robillard v OpenAI Author Class Action in Quebec

Robillard v OpenAI, filed in Quebec Superior Court in September 2025, faced the problem of pirated-input training data for Canadian book authors. The problem was that shadow libraries including The Pile and Books3 pulled Canadian trade books into commercial AI training. The solution Robillard built is a proposed class action that could include thousands of Canadian authors. The suit demands statutory damages of twenty thousand dollars per infringed work and could total several billion dollars. Counsel for the class deployed a certification strategy modelled on Clare v Meta and Bartz v Anthropic.

The impact extends the Canadian AI copyright docket beyond news publishers into trade books and academic monographs, as the pirated books in AI training coverage catalogued across similar US matters. The case has already generated an estimated three hundred million dollars in reserved litigation costs at OpenAI. Robillard raises the pirated-input versus licensed-input bifurcation central to the Bartz settlement in July 2026. The limitation is that class actions in Quebec Superior Court have unpredictable certification outcomes. Certification is scheduled for a hearing window between 2026 and 2027 before Quebec Superior Court. The suit signals that OpenAI faces exposure across every content category Canadian publishers produce.

Common Questions About the Canadian News Lawsuit Against OpenAI

Who are the plaintiffs in the Canadian news outlets lawsuit against OpenAI?

The plaintiff coalition includes Torstar, Postmedia, The Globe and Mail, The Canadian Press, CBC and Radio-Canada, and Metroland Media. These six organizations represent most of Canada's largest national and regional news publishers. They filed jointly in Ontario Superior Court to pool legal costs and present a unified pleading theory against OpenAI.

When was the Canadian coalition lawsuit against OpenAI filed?

The coalition filed the statement of claim on November 29, 2024 in Ontario Superior Court of Justice. The matter is listed as Toronto Star Newspapers Ltd et al v OpenAI Inc et al, file CV-24-00732231-00CL. The case sits on the court's Commercial List, which handles substantial business disputes in the province.

How much in damages does the Canadian lawsuit against OpenAI seek?

The pleading seeks statutory damages up to twenty thousand Canadian dollars per infringed work under section 38.1 of the Copyright Act. Across roughly 16.1 million works, that implies a theoretical ceiling near 322 billion Canadian dollars. The actual award is likely to be substantially lower after judicial calibration under section 38.1(5) of the Copyright Act.

What is the Kimmel ruling in the Canadian OpenAI case?

Justice Kimmel of Ontario Superior Court dismissed OpenAI's motion to strike for lack of jurisdiction on November 7, 2025. The ruling, cited as 2025 ONSC 6217, applied the Van Breda real-and-substantial-connection test to each OpenAI defendant. The decision allowed the case to proceed against six OpenAI entities including OpenAI Inc, OpenAI OpCo LLC, and OpenAI Holdings LLC.

What legal theories does the Canadian coalition rely on?

The pleading advances four principal legal theories to maximize the plaintiffs' recovery exposure for OpenAI. Those theories are copyright infringement under sections 3 and 27, circumvention of technological protection measures under section 41.1, breach of each publisher's terms of use, and unjust enrichment. Each theory maps to independent damages exposure under the Copyright Act and Canadian common law. The circumvention theory does not depend on fair-dealing analysis and runs independent of merits defenses.

How does Canadian fair dealing differ from US fair use for AI training?

Canadian fair dealing is a closed list of enumerated purposes including research, education, criticism, review, and news reporting. US fair use is open-ended and weighs four flexible factors rather than demanding a listed purpose. The structural difference makes Canada less favorable to AI defendants arguing training is a transformative use. Comparative US jurisprudence may inform Canadian doctrine but is not binding on Canadian courts.

Why is the Canadian OpenAI case tied to NYT v OpenAI and Bartz v Anthropic?

All three suits allege unauthorized use of copyrighted content to train commercial generative models like GPT and Claude. The Bartz v Anthropic class action settled for 1.5 billion US dollars in July 2026 across roughly 500,000 titles. The NYT case survived motion to dismiss in March 2025 and continues through discovery before Judge Sidney Stein. The outcomes together shape settlement expectations in the Canadian coalition case.

What is the current procedural status of the Canadian OpenAI lawsuit?

The case is in its discovery phase as of early 2026 after the Kimmel jurisdictional ruling in November 2025. Documentary discovery and oral examinations for discovery are expected to span most of 2026 and early 2027. A trial-ready date is likely between late 2027 and mid 2029 if no settlement is reached before then.

Has OpenAI signed any Canadian publisher licensing deals?

OpenAI announced its first Canadian publisher partnership with Village Media in September 2026, nearly two years after the coalition filing. The deal covers display attribution and includes OpenAI funding for a community navigator product called Open Door. No coalition member has signed a licensing deal with OpenAI as of October 2026. The coalition views litigation as the leverage needed for fair licensing terms.

Which Copyright Act sections are most relevant to AI training in Canada?

Section 3 defines the exclusive reproduction right that publishers assert against AI training. Section 27 characterizes infringement, and section 29 lists the enumerated fair dealing purposes. Section 38.1 sets statutory damages ranges from five hundred to twenty thousand Canadian dollars per work for commercial infringement. Section 41.1 creates civil liability for circumventing technological protection measures like paywalls and bot blockers.

What steps can Canadian publishers take to protect their content from AI scraping?

Publishers should update their robots.txt files to block user-agents used by major AI developers like OpenAI and Anthropic. They should embed machine-readable terms-of-use metadata on every article and deploy anti-scraping defenses from vendors like TollBit or Cloudflare. They should maintain scraping-attempt audit logs and run adversarial prompt tests against ChatGPT. Joining Canadian publisher coalitions pools legal costs and leverage against AI vendors.

What other Canadian AI copyright lawsuits are pending?

La Presse filed a parallel Quebec Superior Court action against OpenAI in November 2025 that mirrors the Ontario coalition theory. VerticalScope sued in Ontario in May 2026 over scraping of its community forum content. Jackson v OpenAI GP LLC is pending in BC Supreme Court as a visual artist copyright matter. Robillard v OpenAI is a Quebec class action for book authors, and the combined docket exceeds fifteen active matters.

How does the Canadian coalition's damages math compare to Bartz v Anthropic?

Bartz v Anthropic settled at roughly three thousand US dollars per work across approximately 500,000 titles. Applied to the Canadian coalition's 16.1 million pleaded works, that per-work rate implies 48 billion US dollars. Even ten percent of that per-work rate still implies a Canadian settlement near 4.8 billion US dollars.

Will the Canadian OpenAI lawsuit influence copyright reform in Canada?

Yes, the outcome is widely expected to drive legislative action on text-and-data-mining exceptions and AI governance. Innovation Science and Economic Development Canada's AI-and-copyright consultation closed in November 2024 with a What We Heard report. A legislative package is expected in the 2027 session that will likely cover training-data law. Rights holders are pushing for consent, credit, and fair compensation while developers want flexibility.

What should readers watch for next in the Canadian OpenAI lawsuit?

Watch for OpenAI's affidavit of documents production in late 2026 and the oral examinations for discovery in early 2027. Watch for any settlement announcements following parallel US matters including NYT v OpenAI. Watch for new Canadian filings and for the next Innovation Science and Economic Development Canada copyright consultation. Appellate rulings from the Ontario Court of Appeal are likely to emerge between 2027 and 2029.