On November 24th, La Presse revealed that it had filed a lawsuit against OpenAI, the company behind ChatGPT, for copyright infringement. We spoke with Patrick Bourbeau, its Vice-President of Legal Affairs, to understand the reasons.
Several Canadian media outlets (CBC/Radio-Canada, The Globe and Mail, La Presse Canadienne, Postmedia, and the Toronto Star) filed a lawsuit in November 2024 before the Superior Court of Justice of Ontario. Why is La Presse filing a lawsuit today, a year later?
Patrick Bourbeau: Quite simply because the statute of limitations in Quebec is three years, whereas it’s only two years in Ontario. ChatGPT 3.5 was launched to the general public on November 30, 2022, so we had until November 30, 2025 to file our lawsuit.
We wanted to take the time to do things properly, hence waiting until the deadline. It does require considerable reflection given that the subject of large language models is relatively new and complex.
What evidence do you have in your possession that would show that OpenAI used your articles for ChatGPT?
Patrick Bourbeau: We have several. First, we conducted tests right from the launch of the platform. We gave prompts to ChatGPT and it responded to us using La Presse content.
We also had an informal study conducted through an American law firm, the one representing the New York Times in its lawsuit against OpenAI as well. It clearly demonstrated that our content was being used by OpenAI.
From that point on, we retained the access logs to our website. And again, we literally see thousands and thousands of visits from OpenAI’s robots. Even after we excluded them with the robots.txt protocol. OpenAI claims that a website editor only needs to write a robots.txt file on their site so their robots don’t retrieve the content. That’s completely false! They don’t respect what they say at all. We can demonstrate this by showing all the logs we’ve collected.
We then set up a sort of wall (editor’s note: a WAF solution in Amazon Web Services) starting September 2, 2025. And even then, OpenAI finds a way to bypass this obstacle by simulating browsers.
Anyone can test it. If you ask ChatGPT “What are the five main news stories on La Presse’s website?”, it will respond.
We made a comparison table between La Presse’s original texts and responses generated by ChatGPT. There are barely a few words that change, but substantively, it’s La Presse’s content reproduced in its entirety.
Did you have any contact with OpenAI, particularly to sketch out settlement attempts as we see in other countries with media outlets that have concluded financial agreements?
Patrick Bourbeau: I can’t speak of settlement attempts, I can’t even call it a discussion. At best, I can call it an attempt at contact. There were a few emails exchanged with OpenAI, but it completely fell through on their end.
To this day, no Canadian media outlet has concluded an agreement with an AI platform. I can’t really give the reason – we can only speculate at this stage – but what we’re seeing is that, for now, these agreements are concluded by platforms on major markets (United States, France, United Kingdom, Germany).
I’ve been told by former employees of these AI platforms that a market like Canada wasn’t even on the radar, as it was too small. So you can imagine the Quebec market with French-language content. There clearly isn’t any interest in concluding agreements on their end.
And I must say that on our end, we don’t have that interest either. That’s not our objective. We don’t want to fall into the same trap that media fell into 20 years ago with the arrival of Facebook and Google. Some concluded partnerships with them and ended up at the mercy of these platforms. We can’t make that mistake again.
What exactly are you asking OpenAI and other AI platforms (Anthropic, Perplexity, etc.) to do?
Patrick Bourbeau: First, you should know that we decided to pursue only OpenAI for financial reasons. Our resources aren’t unlimited – we have to choose our battles. Hence this lawsuit against the biggest player on the market.
The goal isn’t to be compensated through a commercial agreement. The goal is really to be compensated for the use they made of our content in order to build their language model, which contributed to building a company valued at 500 billion dollars today.
Their business model is based on the systematic plundering of content protected by copyright, including La Presse’s content.
Did you have discussions with other Quebec media outlets?
Patrick Bourbeau: There were very preliminary discussions but they didn’t lead anywhere. We couldn’t join the lawsuit of English-language media in Ontario, despite exchanges we had with them at the time, given that the conditions of our website provide that any lawsuit must be filed in Quebec. We couldn’t do it in Ontario – we ran too much risk that it would be dismissed.
What is your broader perspective on this upheaval of generative AI? Does it scare you for your business model?
Patrick Bourbeau: I wouldn’t say it scares us. You can’t be against innovation and we recognize the potential of AI. If the development of these platforms is done legally by compensating content creators for the use of their production, we have no objection.
However, we have no intention of concluding an agreement with an AI platform because we want to maintain a direct relationship with our readers. We don’t want readers to be able to access our content on foreign platforms. Even in the best case scenario, we would lose part of our revenue. We’re betting a lot on the relationship with our readers. That’s what makes the press successful, I think.
Our intention is to develop our own AI tools that we can offer to our readers. I don’t have an announcement to make on this subject but reflection has been initiated on it.
What do you expect now?
Patrick Bourbeau: We’re not the first to file a lawsuit against OpenAI, so we’re starting to see their modus operandi, which seems to be to drag out legal proceedings as much as possible. So we expect quite a long battle.
To date and to our knowledge, OpenAI has never settled a lawsuit out of court. The only platform that has is Anthropic, the maker of Claude, which settled a class action lawsuit for $1.5 billion. We think OpenAI will take at least one or two cases all the way to see what the courts think before making a decision.
Knowing that in the United States, their line of defense is based on the notion of fair use, the equitable use of content.
Patrick Bourbeau: This notion exists in Canadian law but it’s much more restricted. Which means that even if American judgments were unfavorable to content creators, they couldn’t serve as precedent in Canada. Their argument wouldn’t hold water here.
training.isarta.com 
