Privacy, Intellectual Property & Copyright

A key ethical issue around the training and use of Generative AI tools is understanding how data is being used. Privacy, intellectual property and copyright all need to be considered at all stages of Generative AI development and use, including text and data mining (TDM), prompting, and the output of Generative AI tools.

Text and Data Mining (TDM)

As we learned in the first section, Generative AI is trained on massive datasets of texts and images. These datasets may include copyrighted or otherwise protected data or works.

In fact, multiple lawsuits have been filed against Generative AI companies because of suspected unethical use of copyrighted content for training.

• The New York Times has sued OpenAI and Microsoft for using New York Times articles to train their Large Language Models 
• A collective of artists has filed a class-action lawsuit against Stability AI, DeviantArt and Midjourney for using the work of millions of artists as training data 
• Another lawsuit was filed against GitHub Copilot on behalf of open-source programmers who claim that using their code to train the Copilot AI was in violation of the open licenses 

Ongoing discussions in Canada and abroad have focused on whether AI training and TDM require permission from copyright holders or should be an exception to copyright law (Government of Canada, 2025 ; WIPO, 2024 ). These discussions are ongoing as of early 2025.

Prompting

Another question of Intellectual Property is the information and data that is entered into Generative AI tools by users. Different tools have different policies around the collection and use of personal data, including the data required to set up an account and any information included in prompts. Users may inadvertently provide personal or sensitive data in their prompts or queries to AI systems, many of which are saved, stored, and used by companies for training or other purposes.

Data is typically grouped into three risk levels:

• Low risk data is non-sensitive and has minimal impact on privacy. Examples include email addresses and usernames.
• Medium risk data includes personally identifiable information and may present some risk for privacy. Examples include names, addresses, phone number, and birthdays. Note that as more medium risk data is shared simultaneously, it can increase the risk levels (for example, providing a name, birthday, and address is riskier than just providing a name).
• High risk data includes information that has significant implications for privacy risks. Examples include financial information and identification numbers (such as social insurance numbers).

The Freedom of Information and Protection of Privacy Act (FIPPA)  states that you can only use personal information for the purpose it was collection and prohibits the use of an individual’s personal information for other purposes without their consent. The answers in completed assignments, exercises, exams, etc., are considered to be the personal information of the student. (See Western’s statement on FIPPA ).

Before adopting new technologies, it’s important for instructors to review and fully understanding the privacy, security and data management policies and practices of the tool and encourage students to do the same. It’s also important to understand the settings and features of each tool. For example, some tools may have settings that allow you to exclude your prompts from training data or create a temporary chat.

See Western’s Cyber Security Awareness Training for more information on safe digital practices regarding privacy.

Intellectual Property is also an important consideration when using Generative AI. Any information included in your prompts is subject to existing copyright laws. It’s important to ensure that you have permission to use any content (text, images, data, etc.) that you are including in your prompts.

When considering using Generative AI tools to support teaching and learning, it’s especially important to keep in mind that any work created by a student as part of their academic studies is their Intellectual Property and subject to the same copyright laws. (see Policy 7.16 – Intellectual Property ).

Consider the cases below to examine how privacy, copyright and intellectual property might impact the use of Generative AI tools.

Works Generated by AI

Another question of Intellectual Property & Copyright arises with the output generated by Generative AI models. The question that has been raised is whether or not works produced by non-human tools can be copyrighted. Currently, the law in Canada suggests that in order to be copyrighted, a work required significant human contribution (Government of Canada, 2025 ). Works produced with Generative AI may also raise concerns about copyright infringement if they are too similar to existing works.