Carys Craig is a legal scholar and copyright and artificial intelligence expert at Osgoode Hall Law School, York University. She is the director of the Osgoode Professional LLM in Intellectual Property Law Program, and a founding member and the director of Osgoode’s Intellectual Property Law & Technology Program.

What is fair dealing?

Fair dealing is the right to use copyright-protected work as the law allows. The doctrine says it is not an infringement of copyright to deal with original work for research and private study, parody, satire, education, criticism and review, and news reporting. For these purposes, people have the right to use protected works, including by making reproductions that would otherwise infringe copyright.

Several factors determine fairness: the purpose and character of the use, the nature of the work, the amount of use, alternatives, and the effect of the dealing on the work and its market value.

The Supreme Court of Canada has repeatedly said that fair dealing is not just a loophole or an exception. It is central to the copyright system and its purpose: to balance rewarding authors and encouraging the creation and dissemination of works, fostering a vibrant public domain. The copyright system consists of owners’ and users’ rights. Fair dealing is a user’s right, and it is one of the ways that the system achieves balance and avoids defeating its own policy goals.

What is the difference between fair dealing and fair use?

The United States’ fair use doctrine is broader because it is not limited to specific purposes like the Canadian fair dealing doctrine. However, 21 years of the Supreme Court of Canada’s jurisprudence on the scope of fair dealing as a user right has moved it closer to fair use, making the interpretation of its statutory purposes broader and more flexible. For example, the court has said that we need a liberal interpretation of what constitutes research.

There has been controversy about moving to an open-ended fair use defence in Canadian policymaking. CAUT and others have advocated for this shift in copyright law. I have also argued that we must open fair dealing to make it more like fair use. Although, because of the Supreme Court’s jurisprudence, the difference between the two doctrines is increasingly less significant.

How do the two doctrines affect trade relations?

It is typical of the global copyright scene to try to internationalize a high standard of copyright protection for owners with little consideration for the protection of users. Even though fair dealing and fair use are now much aligned, the United States has consistently taken the position in international negotiations that other jurisdictions should not have a broad fair use defence. 

The United States exports and demands strong copyright protections from its trading partners but typically refuses to allow them to develop a broad and flexible fair use doctrine. Some countries have done so, notwithstanding, including Israel and Singapore.

What is the impact of AI on fair dealing and fair use?

There is a retrenchment of copyright control over this paradigm-shifting technology and related resistance to fair dealing protecting AI developers from copyright liability.

Some have argued that the fair use doctrine makes it lawful to create AI training datasets with billions of works for which rights were not cleared, thus safeguarding the AI industry from liability. The courts have yet to settle this question in the United States and Canada. Meanwhile, people concerned about extractive or exploitative use of cultural materials argue that training systems on text and data scraped from the internet is not fair dealing.

There are good reasons to be concerned about the significant effect generative AI could have on our cultural landscape and our educational practices. However, I am wary of the hype around AI that overstates its capabilities and future trajectory. I am equally wary of calls for the law to catch up with, regulate or prohibit this recent technology.

Tell us about the AI-copyright trap

In my forthcoming paper titled “The AI-Copyright Trap,” I caution against strong copyright protection and limited fair dealing as safeguards for creators and culture against the threats posed by generative AI. Relying on copyright to control AI aligns with the corporate interests of the media, cultural and tech industries.

In my keynote address during Fair Dealing Week, I alluded to a backlash against fair dealing. The powerful industry actors calling for copyright law reform in the AI era are the usual suspects who have been advocating to stamp out fair dealing by educational institutions.

I hope people can realize that harming fair dealing and undermining user rights because of generative AI will hurt us in the academic context. For example, the right to engage in research is a fundamental part of fair dealing. This includes the right to engage in informational analysis of large data sets, the right to use AI to extract data and information from original works, and the right to reproduce the works themselves.

We must, therefore, be cautious about the gut response that training AI models on copyright-protected works is unfair because this is the same argument used against educational copying.

Are there risks to academic freedom?

There are risks to academic freedom if there are restrictions on academics to use AI tools or if there is unnecessary monitoring or control by universities or outside interests.

Concerns about generative AI in the classroom have led to efforts to create rules about its acceptable use that might have unintended consequences. In my view, the concepts of academic honesty and plagiarism are sufficient for such purposes.

Similarly, I worry about a future where there is excessive oversight or control over how researchers and academic staff use AI tools, whether it is for setting and grading assignments or for their own writing and research purposes.

Copyright law should be technology-neutral. It is an inappropriate tool to regulate AI and provide guardrails.