Beyond Copyright Battles—AI Art
Last year, copyright drama dominated the conversation. This year, let’s focus on AI art.
Copyright disputes dominated the AI art conversation in 2024.
AI models rely on training data, which raises copyright concerns during their development.
I understand the importance of these discussions. But I’m bored. Same arguments, just different speakers.
My goal in 2025 is to focus on artistic creation and celebrate the pioneers of this new medium.
That’s why I’m starting 👾 Art in Tech.
Understanding the Copyright Battles
Training data shapes what AI models can—and can't—do.
AI models are trained with “publicly available and licensed data”.
Licensed data refers to content explicitly authorized for use.
Publicly available data is content that is accessible online or elsewhere.
Using publicly available data for AI training raises copyright concerns, as there is no consensus on whether public access legitimizes its use for AI training.
Disagreements are around "fair use", which allows using copyrighted material without permission under specific circumstances (e.g., news reporting or research).
Suno and Udio Sued
Udio and Suno are AI platforms that generate music from user inputs.
Major record labels have sued Suno and Udio, alleging they infringe on intellectual property by using copyrighted music to train their models.
Suno and Udio responded that training AI models with copyrighted materials falls under "fair use".
NotebookLM podcast based on RIAA's complaints against Udio and Suno.
NotebookLM podcast based on Suno and Udio response to RIAA's complaints.
Authors Suing OpenAI and Microsoft
OpenAI is facing multiple AI copyright lawsuits.
The New York Times and the Authors Guild, with authors like George R.R. Martin, sued OpenAI and Microsoft for using copyrighted works without permission to train AI models.
NotebookLM podcast based on the Authors Guild complaints against OpenAI and Microsoft.
NotebookLM podcast based on the NYT complaints against OpenAI and Microsoft.
Wrapping up
Don’t get confused. The discussion isn't about whether AI model outputs copy copyrighted works, as that would simply be plagiarism.
The focus is on training data, specifically whether training AI models using copyrighted materials without permission qualifies as "fair use".
And that’s it—all you need need to know on AI copyright.
I’ve only covered two examples, but to visualize every AI copyright lawsuit in the U.S. check out this Wired article.
Focusing on artistic creation
Attending AI art panels feels like an endless loop about copyright issues.
Everything’s been said, and now we are waiting for the courts to decide.
But really, aren’t there AI models trained on copyright-cleared data? Yes!
Stable Audio Open for music and audio generation.
Public Diffusion from Source.Plus for image generation.
Then let’s create art that leaves a lasting, positive impact on history.
Disclaimer. The views expressed are my own and do not reflect the opinions or positions of my employer.