Happy 2026 Biz Talkers!
As we enter a new year, the courts are struggling with a pivotal technology question:
When does the material that AI “learns” from go beyond “fair use” under the copyright law and rise to the level of copyright infringement?
Under the copyright law, the owners of a copyright (whether it is a newspaper, book, website, or other media) have the right to control who copies, performs, displays, distributes, or makes a derivative work of their material. But, it is allowable under copyright law for a 3rd party to use an underlying copyright without needing permission from the copyright holder, so long as the use is a “fair use.”
In the handful of AI infringement court cases decided so far, the issue of fair use has been the primary foundation to determine whether an AI user or company was committing copyright infringement. The AI court showdown appears to be a mixture of the wild west, UFC and WWE.
In one case, the AI output was found to be a “transformative” fair use because it served a different purpose than what was presented by the original copyrighted work. However, in another case, the court said copyright infringement could occur even if the use is transformative if that use is so similar to the copyrighted work that it creates a diluted market by decreasing the public demand for the original work. In another case, a court found infringement when AI created answers to questions based on copyrighted material it was trained on because the AI output not transformative due to creating market dilution of the original work. Lastly, another case is heading to trial because an AI company trained its system on a collection of printed books, and as a result, it was found that the piracy of the books superseded any transformative use.
At this point, there are no written opinions on this issue beyond the trial court level, although there are several matters that are now pending at the appellate level. This area of law is so early in its development that each court ruling must be examined on a case by case basis until the legal dust settles, and a more defined standard for AI copyright infringement is established. So, stay tuned!
In the meantime, let the Olympic law games carry on. But, proceed with caution, particularly if you are the party using AI and you are warranting in your contracts that your AI material is not infringing on the 3rd party copyright it is using or “learning” from.
