Leverage existing laws (Law 25, GDPR)

AI and personal data: Quebec facing algorithmic pillaging

Artificial intelligence is transforming our world at a dizzying pace, but in the shadow of this revolution lies an unknown reality: training an AI model is a technically irreversible act. Once our data is integrated into the mathematical weights of these systems, it becomes impossible to precisely erase its trace. When these models are then distributed as open source, each copy multiplies across the world, making any initial violation permanent and global.

Within a month, Meta will begin ingesting European publications to train its artificial intelligence. In Australia, this is already done, without any real possibility of opposition: family photos, children's faces, artistic creations, everything becomes algorithmic fuel. Why think that Quebec would be spared? Nothing indicates that Meta or other tech giants have respected their legal obligation to obtain explicit consent from Quebecers for this radically new use of their personal data.

Yet, in Quebec, Law 25 is particularly clear and rigorous: any new commercial use of personal information requires manifest, free and informed consent. Training a commercial AI was certainly not part of the initial intentions of users of platforms like Facebook, Instagram, or YouTube. Any collection without consent would therefore constitute a major violation of our legislation. The potential sanctions are considerable: up to 4% of worldwide revenue for serious infractions for companies (more than 7 billion Canadian dollars for Meta alone) and up to 100,000 dollars in personal fines for executives, with a criminal record in Quebec.

Beyond personal data protection, Canadian copyright also protects every work broadcast on these platforms, with compensation that can reach 20,000 dollars per infringement. Considering the millions of Quebec works concerned, the magnitude of the potential damage is staggering.

Even more alarming: several of these companies have already acknowledged using social media data massively. Meta openly exploits Facebook and Instagram content for its LLaMA models. Google has modified its policy to be able to use all public web data, including YouTube. OpenAI has concluded a partnership with Reddit after exploiting its data for years. Elon Musk's xAI draws directly from X/Twitter. Anthropic, while more discreet, invokes "fair use" to justify the use of protected content.

This "fair use" argument, brandished by some companies as a legal shield, however collapses in the face of our legal framework. Law 25 is not interested in data transformation, but in the existence of explicit consent for its use. On this fundamental point, the law is clear: without consent, the infraction is characterized, regardless of whether the data is "transformed" rather than "copied".

I therefore call on Quebec's Commission d'accès à l'information to immediately demand the publication of Privacy Impact Assessments (PIAs) from these companies for their AI projects concerning Quebec data. I demand that all collection be suspended until obtaining the explicit consent required by law. I invite Minister Gilles Bélanger, responsible for Cybersecurity and Digital, and Minister Mathieu Lacombe, responsible for Culture and Communications, to unite their efforts to protect our digital and cultural heritage.

I also solicit all political parties to table a unanimous motion to defend the fundamental rights of Quebecers against this massive extraction of personal and cultural data. Tech giants, who accumulate colossal fortunes thanks to this unbridled exploitation, must understand that no one is above Quebec laws.

The situation is all the more worrying as each trained AI model could constitute a distinct infraction. With Meta, Google, OpenAI and others deploying new versions of their models every few months, the damage and potential penalties multiply at a dizzying pace – possibly more than a hundred billion dollars.

Act, yes, but act now. Each lost day brings us closer to a point of no return. If the CAI lacks resources or hesitates to face these giants alone, then let's create a coalition of citizens, artists, lawyers and experts to support it. Collectively, Quebec has equipped itself with robust law and institutions capable of enforcing it. It's time to demonstrate that our digital sovereignty is not just an abstract concept.

These remarks are made in good faith as part of an essential public debate for the protection of the fundamental rights of Quebec citizens and for the preservation of our cultural heritage in the digital age.

Philippe Bourque, Bs. ENG

Artist, entrepreneur and scientist

Saint-François-de-l'île-d'Orléans

May 6, 2025