Who Owns the Rights to AI-Generated Content? A Gray Area Within Copyright Law

Authors: Paul Déry-Goldberg, Juliette Oger-Chambonnet

The rapid rise of generative artificial intelligence (AI)—particularly language models, such as ChatGPT— is reshaping the way we create. Texts, images, music—in a matter of seconds, these systems can generate creations that, until recently, belonged exclusively to human creativity. However, this innovation raises fundamental questions: Who owns the generated content? And can these creations benefit from legal protection?

Lack of Protection for Content Purely Generated by AI

Copyright law in Canada rests upon three fundamental requirements: a work must be original, fixed in a material form, and—most importantly—created by a human being [1]. A creation generated solely by an algorithm does not meet any of these criteria, according to the Copyright Act.

As a result, content produced entirely through automated processes enters the public domain upon creation. Neither the user nor the model’s designer can claim exclusive rights to it.

AI-Assisted Works: The Nuance That Changes Everything

The legal landscape shifts when a human actively contributes to the creation. In 2021, the Canadian Intellectual Property Office (CIPO) registered a copyright for a pictorial work titled Suryast, co-created by a human artist and the AI application, RAGHAV.

This recognition, though unprecedented, is not universally accepted. In 2024, the Federal Court[2] was petitioned to challenge this registration based on the argument that, under the law, artificial intelligence cannot be considered an author. The outcome of this case could have a lasting impact on how rules governing AI co-created works are interpreted.

Unprotected Works Derived from Protected Data

Although AI-generated content is often not protected by copyright, the data used to train these systems usually is.  Images, texts, audio recordings—the works that feed the models are themselves covered by copyright law. Unauthorized use may therefore render the user or designer of the model liable if it infringes the rights of copyright holders.

A Legal Framework Still Unclear, But Not Nonexistent

Bill C-27, introduced in 2022, would have been considered the first attempt to regulate AI in Canada. It proposed the enactment of the Artificial Intelligence and Data Act, alongside a modernization of the federal personal data protection framework. However, the bill was withdrawn in January 2025 following the prorogation of Parliament. To date, there is still no legislative substitute in place.

However, this does not imply the absence of a legal framework. Laws such as PIPEDA (at the federal level) and the Act respecting the protection of personal information in the private sector (in Quebec) already govern the handling of personal data—including its use in training AI systems. Moreover, several non-binding documents, such as the Principles for Responsible Generative AI published by the Office of the Privacy Commissioner in 2023, provide ethical guidelines for businesses.

In the United States: Legal Status Quo and Executive Withdrawal

Authors highlight that the absence of a federal legislative framework governing artificial intelligence and privacy protection remains a notable issue in the U.S.[3]. It is also reported that President Trump revoked an executive order previously issued by his predecessor, Joe Biden, concerning AI-related risks, citing the need to promote innovation and freedom of expression [4]. Accordingly, the prospect of establishing a unified federal legislative framework is viewed by many as remaining uncertain in the near term.

Best Practices

In this shifting legal landscape, businesses and creators would be wise to adopt cautious practices:

• Transparency: Identify the AI tools being used, especially in commercial or professional contexts.
• Responsibility: Review and correct generated content, as the user remains accountable for it.
• Contractual Vigilance: Read the terms of use for AI tools and define ownership of generated content through contractual agreements.

Need Support?

Determining the applicable rights for content generated by artificial intelligence can be complex, especially within a constantly evolving legal framework. Whether you are using AI to create content, train models, develop products, or establish partnerships, our team is here to help. We assist with assessing legal risks, identifying rights holders, drafting tailored contractual clauses, and implementing responsible and compliant practices. Whether you are a tech company, a creator, or an adopter of AI solutions, we are here to provide proactive and secure guidance.

[1] Copyright Act (R.S.C., 1985, c. C-42), section 5(1); CCH Canadian Ltd. v. Law Society of Upper Canada, [2004] 1 S.C.R. 339, 2004 SCC 13.

[2] Samuelson-Glushko Canadian Internet Policy and Public Interest Clinic (CIPPIC) v Sahni, July 8, 2024, Federal Court, file no. T-1717-24.

[3] Joshua Turner, Nicol Turner Lee, “Can California fill the federal void on frontier AI regulation?”, Brookings, June 4, 2024, online.

[4] “INITIAL RESCISSIONS OF HARMFUL EXECUTIVE ORDERS AND ACTIONS”. The White House, January 20, 2025, online: (https://www.whitehouse.gov/presidential-actions/2025/01/initial-rescissions-of-harmful-executive-orders-and-actions/)