Your Independent Journal from the Heartland
Artificial intelligence programs can be very useful, but you have to be aware of how it often infringes on copyrights and it sometimes makes up sources. When AI searches copyrighted materials, such as books, newspapers and magazines, copyright infringement is very likely. There have been more than 30 federal lawsuits against AI programmers alleging copyright infringement.
The AI companies have argued, mostly unsuccessfully, that the Fair Use Doctrine protects their use of copyrighted materials. Lynn Goldsmith, a professional photographer, took a photo of the musician Prince. Andy Warhol used this photo to create a silkscreen print. The U.S. Supreme Court ruled that it was not fair use for Mr. Warhol to adapt Ms. Goldsmith’s photograph. Andy Warhol Foundation for the Visual Arts, Inc. v. Goldsmith, 598 U.S. 508 (2023).
The copyright laws do not define “fair use” in terms of how many words from a book or article can be copied. However, the International Association of Scientific, Technical & Medical Publishers suggests that no more than 500 words from a work can be used, and if it is used, it must be attributed to the author. AI can be very useful in creating scientific and medical articles, but humans should check citations to make sure that fair use concepts are not violated and that the citations used are accurate.
There are three recent cases that give insights on artificial intelligence: Andersen v. Stability AI, New York Times v. Open AI, and Getty v. Stable Diffusion. In Andersen v. Stability AI, U.S. District Judge William Orrick (Northern District, California) denied Stability’s Journey’s motion to dismiss the artists’ copyright infringement claims, allowing the case to move towards discovery and trial. The judge found both direct and induced copyright infringement claims to be plausible. The induced infringement claim against Stability AI argued that by distributing their model “Stable Diffusion” to other AI providers, the company facilitated the copying of copyrighted material.
In New York Times v. Open AI, the core allegation is that Open AI is infringing on copyrights through the unlicensed and unauthorized use and reproduction of Times works during the training of its models. U.S. District Judge Sidney Stein (Southern District New York), whose order rejected parts of Open AI motion to dismiss, ruled that the Times’ “numerous” and “widely publicized” examples of ChatGPT producing material from its articles justified allowing the claims to continue.
Getty, an image licensing service, filed a lawsuit against the creators of Stable Diffusion in England, alleging the improper use of its photos, both violating copyright and trademark rights it has in its watermarked photograph collection. The trial took place in June in London.
These three cases demonstrate that using AI to create material is risky business and that many AI programs violate the copyright laws of the United States and other countries. Defending a copyright infringement lawsuit is expensive and time-consuming.
Concerning legal and medical research, some AI programs have made up articles and citations. In addition to refercing studies that do not exist, Health Secretary Robert F Kennedy Jr.’s “Make America Healthy Again” report appears to have used garbled artificial intelligence to generate scientific citations, in addition to referencing studies that do not exist. A NOTUS investigation found seven studies referenced in Kennedy’s 68-page report that were either wildly misinterpreted or never occurred at all. NOTUS contacted the cited authors of these seven studies and confirmed that they diid not support the report’s findings. NOTUS is a new journalism organization that covers Washington, D.C. with a nonpartisan perspective.
P. Kevin Castel, a New York federal judge, sanctioned lawyers who submitted a legal brief written by the artificial intelligence tool ChatGPT. The brief included citations of non-existent court cases. In addition to each paying a $5,000 fine, the attorneys, Peter LoDuca and Steven Schwartz, and their Levidow law firm, Judge Castel criticized the lawyers harshly and ordered them to send a copy of his opinion to each of the real-life judges whose names appeared in the fictitious filing. Mata v. Avianca, Inc. (S.D.N.Y. 2023).
If you use AI to produce a report, write an article or a book, you must review the material carefully. Every citation must be double-checked for accuracy. In addition, if parts of other copyrighted works are copied, make sure that they are brief and properly footnoted with an accurate source reference. Lawyers should not rely on AI to write their briefs. By using AI lawyers risk sanctions and may waste more time double-checking AI language than it would take to write the brief themselves.
Joel D. Joseph of La Jolla, Calif., is a lawyer who has practiced copyright law for 40 years. Email joeldjoseph@gmail.com
