Copyright, which pertains to the "right to copy", gives copyright owners the right to prevent others from reproducing their work or copying any substantial portion of it. Owning the copyright for a work means you have the exclusive right to commercially benefit from its use. Additionally, copyright grants moral rights which include attribution and the right to modify a work. However, copyright is not a property right. Copyright does not protect ideas, but the specific expression of ideas.
Fair dealing is the user's right to make use of copyrighted works. BCIT has published Fair Dealing Guidelines. Fair dealing protects users who use part of a work for an allowable purpose such as education, research, private study, satire, parody, criticism, review or news reporting. In Canada, fair dealing only applies to these purposes, and dealing for a commercial purpose weighs against a fair dealing finding. As an emerging technology, GenAI is also an emerging area of law, and it will take years for lawsuits to work through Canadian courts and provide clear copyright precedent. As of February 2024, no Canadian lawsuit has been filed against a GenAI company.
Lawsuits have been filed in the US, where most GenAI companies are based. In the US, the user's right to copyrighted work is called fair use. It is similar to Canada's fair dealing, but fair use does not need to fall under a specific allowable purpose. There is even precedent for companies being able to use copyrighted materials for commercial purposes without permission so long as it is a "transformative" use. GPT stands for Generative Pre-Trained Transformer, a hint that GenAI companies fundamentally claim to fall under this fair use exemption
Such uses, if found to be fair use, can be done without permission from the copyright owner. GenAI companies use vast quantities of copyrighted data without permission to train their models. The biggest source of data is Common Crawl, a collection of more than 250 billion publicly available webpages collected and copied through a practice called web scraping. If such uses are found to not be fair use, with GenAI companies like Google and Microsoft valued in the trillions of dollars, GenAI may turn out to be the biggest copyright infringement in history.
It depends. In Canada, current requirements for copyright protection include originality, fixation, and the exercise of skill and judgement. These requirements can be understood to mean that content created by GenAI is not protected unless an original work is created with the sufficient addition of human skill and judgement.
In October 2023, the Government of Canada launched a consultation on the implications of generative artificial intelligence for copyright. The result of this consultation may be new legislation about GenAI and copyright in the form of an amendment to the Copyright Act.
In the United States, GenAI output cannot be copyrighted unless a human shows "sufficient creative control" over the process. The Copyright Office has refused a copyright application for an image autonomously created by an algorithm. In a separate application, copyright was partially granted for a graphic novel which contained human-authored text and AI-generated images. The human-authored text and human selection and arrangement of text and images was protected, but not the AI-generated images themselves.
While there is nothing preventing a copyright owner from filing a copyright infringement lawsuit, even frivolously, some GenAI companies have started to offer legal indemnity to their users. In practice, this drives lawsuits their way, and gives them control over the legal defences employed in precedent-setting legal cases. Further, it deters copyright owners, particularly individuals, from deciding to go against a Big Tech company with a massive legal team.
Google wrote "If you are challenged on copyright grounds, we will assume responsibility for the potential legal risks involved."
OpenAI Copyright Shield: "We will now step in and defend our customers, and pay the costs incurred, if you face legal claims around copyright infringement".
Anthropic has stated "We will defend our customers from any copyright infringement claim".
Microsoft Copilot Copyright Commitment: "We believe in standing behind our customers when they use our products."
There are hundreds of billions of investment dollars at stake in these companies, and copyright infringement lawsuits are a threat to their very existence. While you still might get sued, your legal bills may be covered, at least for these companies. GenAI surveillance companies are another story.
"AI text detection" products purport to detect the use of ChatGPT or other GenAI tools. These have been widely found to have an unacceptably high rate of false positives, which cause harm to students.
In 2023, BCIT's Educational Technology and Learning Design Committee voted to discontinue the AI detection functionality in Turnitin. There is no approved AI Detection tool at BCIT and they should not be used.
No AI Detection tool (such as GPTZero, Copyleaks, and ZeroGPT) has been evaluated for compliance with BCIT policies. Uploading student personal information to an unvetted service may be a breach of British Columbia's Freedom of Information and Protection of Privacy Act. Additionally, it may be a violation of the Copyright Act as students own copyright of their work.
It's wrong to upload student copyrighted work to a third-party who may exploit it in unlawful ways. There's a saying, "if you're not paying for it, you're the product". It is unethical to pay for academic surveillance technology with student data.
The full talk description and other talks in the series by Dr. Brenna Clarke Gray, Rebecca Sweetman, and Dr. Ann Gagné can be found on UNBC's AI Literacy LibGuide.
Teaching with Generative AI LibGuide by BCIT Library Services is licensed CC BY-NC, meaning it can be used for non-commercial purposes if attribution is provided. Learn more about Creative Commons licenses on the BCIT Open Education LibGuide.
.