Writers vs. AI vs. the Copyright Act (Research Essay)
Please Note: This research essay was originally written for my Copyright and Contracts class. It was originally written in MLA. Formatting has been changed slightly to align with the style of the website.
Word Count: 3061
Copyright within Canada is often a contested issue, due to the fact that governing laws and acts such as the Canadian Copyright Act (the Act) lack much needed updated and modern language. As technology, such as Artificial Intelligence (AI), becomes ever more present in today’s literary world, it is more imperative now than ever to put the proper policies in place to protect this country’s creatives and artists. According to Anja Karadeglija, AI can create a number of digital products that range from images, text, videos and even computer code (Karadeglija).
In 2023, the Government of Canada released a consultation titled “Consultation On Copyright In the Age of Generative Artificial Intelligence”. In this document the Government acknowledges the issues that creatives face as a result of AI’s existence in relation to how the Act lacks the language to define it. They also acknowledge the value that the creative industries contributed to the Canadian economy in the year of 2022, the year prior to when this consultation was released. Unbeknownst to many Canadians, the Government invested a substantial amount of money into the movement of AI as far back as 2017. This shows that as of now at least, the Government makes no plans to eradicate it from our culture and online experience. This is further exemplified by the fact that they have increased spending as recently as 2021 according to the consultation they have released (“Consultation Paper”).
There are even instances where AI has managed to make fairly disruptive moves in the online space. This includes an instance where a man from India managed to have his work registered for copyright with the Canadian Intellectual Property Office (CIPO). To the surprise of many, it worked. Especially when one looked further into the case and realized this painting named Suryast was denied copyright registration in almost every other country that he applied for. This was due to the fact that other countries’ copyright offices deemed that not enough human intervention and skill were applied to the work. Another instance where the Act and copyright went “head to head” is with OpenAI, the creator of Chat GPT, and multiple Canadian outlets that filed a claim against them. These Canadian outlets that joined together in a civil claim, accused OpenAI of scraping their databases for content without their permission. Both of these cases exemplify how copyright within the Act is considered outdated. There are no clauses that specifically mention AI and its dubious connections to possibly stealing others’ works.
Due to the fact that the Canadian Government has invested a large amount of money and time in the AI industry, it is unfortunate to say that this type of technology will not be going anywhere anytime soon. The best solution at this point in time involves creatives unifying in an effort to better protect their work against bigger tech companies ie., companies that are able to get away with taking others’ works without permission and eventually profiting off of it. This is an ongoing issue that, without the help of our government, will never go away.
Firstly, the Copyright Act (the Act) is outdated in a number of different ways. For example, there is a final part of the act that outlines existing rights. “In the case of an essay, article or portion forming part of and first published in a review, magazine or other periodical or work…the right shall be subject to any right of publishing the essay, article or portion in a separate form to which the author is entitled on January 1st, 1924…” (Copyright Act, 168). This part of Schedule I found on page 168 of the PDF version of the act, details the existing rights a person may have if their work was created on or before January 1st 1924. This date was well over a hundred years ago, and yet it is still included in the Act. In order for someone to be covered by this, they would have to be born in approximately 1905 or earlier and subsequently created a work and have been a legal adult when that work was created.
According to an article by Moira Donovan titled, “Play Catch-Up On Copyright” from the National, subsections of the copyright such as “Temporary Reproductions for Technological Processes” and “Fair Dealings” play a hand in allowing companies like OpenAI to get away with their business practices (Donovan).
As per the Act, Temporary Reproductions for the Technological Processes defines technological parts of a process not infringing on copyright as long as it is an essential part of the process (Copyright Act, 67). It also states that infringement does not take place as long as the reproduction only takes place during the the duration of the technological process (Copyright Act, 67). The issues with these clauses respectively, is that they are unable to determine where the technological processes involved with creating AI fall into this act. As far as the user base and creatives, who believe their work is being unjustly used, are concerned; there is no end to the technological process of AI. In other words the “duration of the technological process” is infinite. Or it will run until the company discontinues their services.
Fair Dealing on the other hand presents an entirely different issue when it comes to the Act. For example, in section 29 of the Act it reads, “Fair dealing for the purpose of research, private study, education, parody or satire does not infringe copyright” (Copyright Act, 43). Generative Artificial Intelligence can be defended on almost every aspect of this clause in particular. AI can be used for research, meaning, a user can ask a model such as Chat GPT to explain how books are made. The model will pull information from any website it was fed that had such detail on the process. A user can receive needed information on a number of different subjects based on whichever websites the machine has learned from, and that can legally be called private study and education. Both of which, according to the Act, fall under Fair Dealing.
These clauses, respectively, should be redone in such a way that ensures copyright holders of the works, that are used to train these machines, are at the very least compensated for the works use. An amendment to the Fair Dealing clause should look something like this:
29 Fair dealing for the purpose of research, private study, education, parody or satire does not infringe copyright unless the following happens:
generative artificial intelligence is used and both the user and maker of the model cannot verify jointly where the resulting work got its inspiration from;
(i) the user and maker cannot verify if the natural persons involved in making the inspiring work gave explicit permission to have their work be used;
(ii) the user and maker of the model make any profit off of the work resulting from the use of the model without the knowledge and consent of the creators responsible for the conception of the inspiring works used.
Everything from “unless the following happens” and onwards is not currently a part of the Act. However, for the sake of creatives and their potential loss of income, due to the nature of generative AI and the composition of its models, this can be seen as a jumping off point.
“...generative artificial intelligence is used and both the user and maker of the model cannot verify jointly where the resulting work got its inspiration from….” would mean that both the users of the AI model and the makers need to be able to verify where and who the resulting work came from. “..the user and maker cannot verify if the natural persons involved in making the inspiring work gave explicit permission to have their work be used….” means that infringement takes place even if the user and maker of the model can verify where the resulting work came from, because the creator of the inspiring work did not consent to have their work be used in that way. Lastly, “...the user and maker of the model make any profit off of the work resulting from the use of the model without the knowledge and consent of the creators responsible for the conception of the inspiring works used….” means that even the user and maker know where the inspiring work came, the creator of the inspired work was informed; if their informed consent is not given, under no circumstances can the maker or the user make any profit off of the work.
In an article by Dan Milmo, he extracted a quote from the Telegraph of OpenAI saying that in order to properly train AI models they would have to use works outside of the public domain. “Because copyright today covers virtually every sort of human expression - including blogposts, photographs, forum posts, scraps of software code, and government documents - it would be impossible to train today’s leading AI models without using copyrighted materials” (qtd. in Milmo). This quote shows on a surface level why individuals behind these companies want their models to have continuing access to various copyrighted works. As of December 5th, 2024 Open AI was valued at approximately 157 billion dollars according to Moira Donovan’s article (Donovan). Thus showing how lucrative and attractive this business model is for those working at the heads of companies like OpenAI.
There are specific cases that show, allowing AI based works to be copyrighted in the first place can harm others. For instance, a man by the name of Mr. Ankit Sahni used the AI model RAGHAV Artificial Intelligence Painting App to create a painting called Suryast. Suryast was inspired by the Starry Night, a painting that was originally done by Vincent Van Gough in the late 1800s (“Vincent van Gogh. The Starry Night”).
In July of 2024, the Canadian Internet Policy and Public Interest Clinic (CIPPIC), filed an application to the Canadian Federal Court to have Mr. Sahni’s copyright removed (Medeiros, Maya, et al). They cited that Mr Sahni’s piece did not have enough human intervention to constitute having copyright law apply to it. According to the article by Medeiros, Maya, et al; CIPPIC made two arguments in their application. “...1) Suryast does not meet the originality requirement for copyright; and 2) an AI system cannot be an “author” under the Copyright Act” (Medeiros, Maya, et al).
Generative AI, such as OpenAI and RAGHAV, involve text and data mining, according to the consultation by the Canadian Government (Government of Canada). It needs to be trained based on works done and performed by real life people. CIPPIC argues that those grounds and the ones set out by the Act, no original work can be created as a result of its use. The basis for CIPPIC’s second argument is that AI models cannot be authors of a work, since they are not a “natural person”, which in this case refers to an author and therefore, a human being. “...an AI system cannot exercise the common intent required for joint authorship” (Medeiros, Maya, et al).
The Medeiros, Maya et al. article also cites Mr. Sahni’s experience trying to have the work registered with the US Copyright Office, where they refused a second appeal. But according to the Canadian Copyright Database, the copyright is still in place for Mr. Sahni in Canada (“Canadian Copyright Database”).
Another instance where AI has been an issue with Canadian copyright is of course with OpenAI. “Playing Catch-Up On Copyright” by Moira Donovan reports that “the CBC, the Canadian Press, the Globe and Mail, Post Media, the Toronto Star, and Metroland, filed their statement of claim to the Ontario Superior Court of Justice last week” (Donovan). These outlets claim that OpenAI has used their articles to train their AI model ChatGPT without their permissions. The claim was filed at the end of 2024. However, this article reaffirms that Canadian copyright law does not address these types of concerns regarding AI and its influence. It also addresses issues where the AI model does not directly copy a specific work from these outlets. “Kerr-Wilson says it gets trickier in situations where the output was heavily influenced by the content used in training but isn’t an exact copy. This appears to be the case in the Canadian lawsuit, as the claim doesn’t include examples of verbatim copying” (Donovan). Whereas, in a case with the New York Times suing OpenAI, they were able to prove that verbatim copying took place. OpenAI responded saying that they were working on their system so that it would no longer regurgitate almost exact copies of existing works. This illustrates how the holes in the Act will have an impact on how creative works are utilized by these models. Specifically, it will impact individuals like Mr. Sahni and their ability to have their work registered with the Canadian Intellectual Property Office. As well as their ability to make a profit off of the work they “create” with AI.
In their consultation, the Canadian Government noted that the creative industries as a whole contributed over 60 billion dollars to the Canadian Economy in 2022. In a turn of events, AI stakeholders are also worried about how the Act will impact this section of the technology industry. They worry, more specifically, that the lack of definition will slow down investments over the years. As mentioned earlier, the Government, as far back as 2017, invested 125 million dollars into industry. The consultation does not say where to whom. But they say that in 2021, they announced they would invest 443.8 million dollars to businesses such as Scale AI which hopes to build supply chains in Canada that are AI-powered (Government of Canada).
In this consultation, the Government consulted multiple stakeholders in various industries that could be impacted by the lack of direction from the Act. A mutual concern amongst many was whether or not creators of AI needed permission from the copyright works they took from via text and data mining. Stakeholders from the technology sector argued that permission was not required. Opposingly, stakeholders from the creative sector argued that, at the very least, rights holders should be compensated for their work being used (Government of Canada). However, both sides agreed that it was too early to make any definitive stances on how AI should be addressed in terms of ownership and authorship of a work.
In June 2022, the Government introduced Bill C-27 to parliament which is otherwise known as the Artificial Intelligence and Data Act (AIDA). The consultation says this act will enforce responsible development of AI systems. Unfortunately, it is not all encompassing as creatives would hope for. “...its scope is designed to complement existing legal frameworks, such as copyright, and is thus not the vehicle to address many concerns about the impact of AI on creative content” (Government of Canada). It can be argued that the Copyright Act should be amended on its own instead of introducing a new bill. According to the House of Commons website there is something called the “parent act rule”. This rule means that in order to amend an existing act (also known as the parent act), a new bill has to be very specific about how it introduces any new legislation. “In general, any amendment to a section of the parent Act not otherwise amended by the bill is inadmissible; any amendment to an Act not mentioned in the bill is inadmissible” (“Amending Bills at Committee and Report Stages”). Regardless of how it is done, the process is relatively the same.
Forty-eight hours written notice for any public bill is required. After it appears on the Notice Paper it will then go to the Order Paper which then means it is ready to be introduced to the House. Then it must go through the second reading where the overall contents of the bill are explored. It will subsequently be referred to a committee where witnesses are invited to provide their comments on the bill. Then the report stage takes place, this is when members of the House can present any amendments to bills they think are necessary. The third reading covers the final version of the bill before it goes to the senate. Then the senate must debate over the bill in a process that is similar to the House of Commons. If the Senate has no amendments to make, then the bill will reach Royal Assent. Which means the Crown’s representative, the Governor General, will approve the bill in which case it will become law. In order for Bill C-27 to pass, it will have to go through the above mentioned process.
According to the “The Artificial Intelligence and Data Act (AIDA) – Companion Document”, the bill has yet to pass even though it was introduced in 2022. It has been almost three years since, and the bill is still being debated.
Finally, generative artificial intelligence is a technological process that the Canadian Government is heavily invested in. It is no secret that by now the creative industries in Canada have an increasing concern over the boundless abilities of AI. It includes mining content from all over the internet without the permission of the people it takes from. The lack of revisions in the Copyright Act are noticeable and in the view of these creatives is causing more harm than good. Proposed legislation does not include, as of right now, any specific amendments to the Copyright Act itself. Since AI has been heavily invested in by our government, the most that creatives can hope for is amendments to the act or new bills that better protect them from AI continuing current practices.
Cases such as Mr. Ankit Sahni are relevant in how the Canadian Government is possibly biased in their ruling of the granted copyright of Suryast. Organizations such as CIPPIC are doing their best to elevate the voice of creatives and their works, in order to make sure that they are receiving the dues that are owed to them. While they have yet to win against Mr. Sahni, they are still continuing in their mission by providing help to those affected by discrepancies in the Copyright Act.
As of now, organizations like CIPPIC, and like-minded individuals, can hope that acts like AIDA and Bill C - 27, will add to the sacredness of creative works and how they are protected by the Copyright Act.
Works Cited:
“About Us.” Cippic.ca, 2023, www.cippic.ca/about-us. Accessed 29 Mar. 2025.
“Amending Bills at Committee and Report Stages .” Ourcommons.ca, 2025, www.ourcommons.ca/procedure/guides/amendingBills-e.html. Accessed 30 Mar. 2025.
“The Artificial Intelligence and Data Act (AIDA) – Companion Document.” Canada.ca, 31 Jan. 2025, ised-isde.canada.ca/site/innovation-better-canada/en/artificial-intelligence-and-data-act-aida-companion-document. Accessed 30 Mar. 2025.
Cahill, Joe. “From Original Photo to AI Art: SURYAST Lacks the Required Human Touch for Copyright Protection.” Passle, 22 Dec. 2023, ourtake.bakerbotts.com/post/102ivxv/from-original-photo-to-ai-art-suryast-lacks-the-required-human-touch-for-copyrig. Accessed 3 Apr. 2025.
“Canadian Copyright Database.” Ic.gc.ca, 2021, www.ic.gc.ca/app/opic-cipo/cpyrghts/dtls.do?fileNum=1188619&type=1&lang=eng. Accessed 30 Mar. 2025.
“Consultation Paper: Consultation on Copyright in the Age of Generative Artificial Intelligence.” Canada.ca, 5 Dec. 2024, ised-isde.canada.ca/site/strategic-policy-sector/en/marketplace-framework-policy/consultation-paper-consultation-copyright-age-generative-artificial-intelligence. Accessed 3 Apr. 2025.
Copyright Act, RSC, 1985, c C-42.
Donovan, Moira. “Playing Catch-up on Copyright.” Nationalmagazine.ca, 2024, nationalmagazine.ca/en-ca/articles/law/hot-topics-in-law/2024/playing-catch-up-on-copyright. Accessed 16 Mar. 2025.
Graves, Franklin. “Copyright Office Affirms Its Fourth Refusal to Register Generative AI Work.” IPWatchdog, 12 Dec. 2023, ipwatchdog.com/2023/12/12/copyright-office-affirms-fourth-refusal-register-generative-ai-work/id=170564/. Accessed 3 Apr. 2025.
Government of Canada. “Copyright.” Canada.ca, 11 Feb. 2025, ised-isde.canada.ca/site/canadian-intellectual-property-office/en/copyright. Accessed 30 Mar. 2025.
“How a Government Bill Becomes Law.” Queensu.ca, 2024, guides.library.queensu.ca/gov/canada/federal/how-a-bill-becomes-a-law. Accessed 30 Mar. 2025.
Karadeglija, Anja. “Canada, Other Countries Struggling to Come up with New Rules for AI and Copyright.” CTVNews, 2 Mar. 2025, www.ctvnews.ca/sci-tech/article/canada-other-countries-struggling-to-come-up-with-new-rules-for-ai-and-copyright/. Accessed 29 Mar. 2025.
“Legislative Process - Our Procedure - ProceduralInfo - House of Commons of Canada.” Ourcommons.ca, 2017, www.ourcommons.ca/procedure/our-procedure/LegislativeProcess/c_g_legislativeprocess-e.html. Accessed 30 Mar. 2025.
Medeiros, Maya, et al. “Can AI Be an Author? Federal Court Asked to Decide in New Copyright Case.” Nortonrosefulbright.com, 6 Aug. 2024, www.nortonrosefulbright.com/en/knowledge/publications/ad12aba2/can-ai-be-an-author-federal-court-asked-to-decide-in-new-copyright-case. Accessed 16 Mar. 2025.
Milmo, Dan. “‘Impossible’ to Create AI Tools like ChatGPT without Copyrighted Material, OpenAI Says.” The Guardian, 8 Jan. 2024, www.theguardian.com/technology/2024/jan/08/ai-tools-chatgpt-copyrighted-material-openai. Accessed 16 Mar. 2025.
Paas-Lang, Christian. “Bills Now Take Almost Three Times as Long to Get through the Senate.” Nationalpost, National Post, 30 June 2019, nationalpost.com/pmn/news-pmn/canada-news-pmn/bills-now-take-almost-three-times-as-long-to-get-through-the-senate. Accessed 30 Mar. 2025.
“Vincent Van Gogh. The Starry Night. Saint Rémy, June 1889.” The Museum of Modern Art, MoMA, 2015, www.moma.org/collection/works/79802. Accessed 30 Mar. 2025.
