AI Tools and Resources

When using online or library resources in generative AI projects, it’s crucial to understand their ownership and licensing rules. Access through USF Libraries or tools like Google Scholar may include content governed by specific agreements, so verify usage rights before incorporating materials into your project.

Here's what you need to know:

1. Licensing Restrictions

Your authentication with USF systems allows you to seamlessly access Libraries negotiated electronic resources through online searches like Google Scholar.  Most electronic resources accessed through USF Libraries are governed by license agreements. These licenses may

• Restrict use to USF students, faculty, and staff;
• Permit use only for non-commercial, educational, teaching, and research purposes;
• Prohibit commercial use, systematic downloading, and automated data scraping (e.g., using bots or scripts).

Using licensed content to train or feed into generative AI tools—especially if the tools are commercial or publicly accessible—may violate these agreements, even if the intent is educational.

Non-Libraries’ resources may also come with licensing restrictions.  If you access content online via a user log-in, like on a social media platform, your use of that content must conform to the platform terms and conditions.

Check the terms of library subscription resources and best practices for text and data mining before beginning your project.

2. Copyright and Derivative Works

Even if a resource is accessible through the library, it may still be protected by copyright. Generative AI projects that remix, transform, or repurpose content could be creating works that would be considered derivative. Creating a derivative of a copyright protected work typically requires permission from the copyright holder unless an exemption, like fair use, could apply.

3. Fair Use Considerations

Fair use is an exception in copyright law that describes certain circumstances where the use of copyrighted works is not an infringement.  Fair use must be considered on a case by case basis and consider four factors:

• The purpose of the use (e.g., nonprofit educational vs. commercial).
• The nature of the copyrighted work.
• The amount used.
• The effect on the market for the original work.

Using large volumes of copyrighted material to train AI models may weigh against fair use.

4. Best Practices

• Check license terms or contact your subject librarian before using library materials in AI projects.
• Avoid using entire works or large datasets unless explicitly permitted.
• Consult with the Copyright and Intellectual Property Librarian for guidance on specific use cases.
• When in doubt, seek permission or use openly licensed or public domain materials.

Check these other resources for more information: