What Is Fair Use?
We recently covered the Copyright Office’s report on AI training and fair use. Fair Use is a part of copyright law that allows limited use of copyrighted materials. It applies when usage occurs under certain conditions, without the copyright holder’s permission.
How AI Training Works
AI “training” uses data, including copyrighted works, for the AI to learn patterns. This data teaches models how to generate new text, images, or other outputs.
The Meta Lawsuit
Thirteen authors sued Meta for copyright infringement. The judge ruled the plaintiffs failed to show that Meta’s outputs were substantially similar to their works.
He noted that copying protected works without permission is generally illegal. However, the plaintiffs did not prove that Meta’s use caused market harm.
Why the Court Found Fair Use
In this case, using copyrighted material was considered fair use. This decision was based only on insufficient evidence of market harm. It does not settle the broader question of AI training legality.
Remaining Claims Against Meta
A separate claim alleges Meta unlawfully distributed copyrighted works. That claim was not dismissed and will continue through the courts.
Judge’s Comments on Meta’s Defense
The judge rejected Meta’s public interest argument for using copyrighted material. He called Meta’s claim that blocking the works would halt AI development “nonsense.” He added that if the books were truly essential, Meta should have licensed them.
What This Means for the Future
Although Meta won this case, stronger claims or better evidence could change outcomes. This ruling does not provide blanket permission for AI companies to use copyrighted works.
Other articles on LLM training and Fair Use: Copyright Office Report on AI Training and Fair Use
AI and Copyright: Court Rules Against Fair Use in Training AI Models