13 Copyright, intellectual property, and GenAI
Who owns what Generative AI creates?
When you use Generative AI (GenAI), there are some serious concerns about ethics, laws, and copyright. It’s really important to pay close attention to these, especially in your coursework.
The Copyright Librarian from University Libraries has created important information about copyright and AI:
- What GenAI Learns From: AI tools learn from huge amounts of information, called “training data.” We generally don’t know what information is in a GenAI tool’s training data, and if it includes copyrighted material. We don’t know if the copyright holder consented to having their content used to train the GenAI.
- What GenAI Creates: This also brings up questions about who actually owns new content that GenAI creates based on its training data (for example, when you give it a prompt and it generates a response).
- Copyright in the US: In the United States, for something to be protected by copyright, a human being must have created it. This means that things GenAI creates on its own cannot be copyrighted.
- When You Add Your Touch: However, if you take something GenAI creates and then add your own creative changes or expressions to it, your changes can be copyrighted. Many people adjust what GenAI produces, and those human-made changes can then be likely protected by copyright. You can find more details at the University Libraries Copyright website.
definition
The dataset used to teach an AI model.
