IS THE PRODUCT OF AN ARTIFICIAL INTELLIGENCE ALGORITHM “SPEECH” AND ENTITLED TO PROTECTION UNDER THE FIRST AMENDMENT?

This question is certain to arise in many contexts but is particularly relevant in the area of copyright law.  The laws of copyright were written to protect human creative content in books, songs, photographs, etc.  A common defense to copyright infringement, (of a song for example), is that the offending product is a new “work” and is therefore protected “speech” under the First Amendment and/or permitted by the “fair use” doctrine.  But is it speech at all if it was created by a machine? Must protected speech emanate from a human brain?

There is little doubt the framers of the constitution were thinking of human speech when they adopted the First Amendment.  Even Benjamin Franklin, who was likely responsible for including in the document the protection of intellectual property, (Article I, section 8) did not anticipate AI and machine manufactured speech.

The courts have held that many types of entertainment are speech.  Movies, books, songs, photographs, even video games. The copyright laws and statutes were designed to protect such creative works but also to reach a balance between such protection and fostering the development of new ones.

The First Amendment and the fair use doctrine codified in the Copyright Act, (17 U.S.C. Sec. 107) provide the protection needed for new authors to express their creativity.  But the entire assumption behind this protection is that the new work is a form of speech.  It follows then that if the new work is not “speech” then it is not protected by the Constitution or the statute.

The U.S. Supreme Court addressed the balancing of these two interests earlier this year in a case involving a copyrighted photograph and a subsequent silk screen work prepared by Andy Warhol.  The Court noted that to avail oneself of the protection provided by the First Amendment through the fair use doctrine, the subsequent work must be “transformative”. The Court said that meant the new work must reflect, “a further purpose or different character” and one that was, “sufficiently distinct from the original.” Andy Warhol Found. for the Visual Arts, Inc. v. Goldsmith, 143 S.Ct. 1258 , 1277, 1287 (2023).  The court found that Mr. Warhol’s subsequent work did not meet that standard and constituted copyright infringement.

Presumably the AI produced work could be programmed to satisfy this transformative standard, but it would be irrelevant if the court found that the work was not speech to begin with.  So, what is “speech” anyway? The Supreme Court has applied it broadly to talking, writing, printing, broadcasting, entertainment endeavors and other forms of expression.  It also applies to symbolic expressions such as displaying or burning the flag, all of which are distinctly human endeavors.

AI developers have begun using copyrighted works, (books in particular) to train their creations. The results are prose that mimic the original and may violate the rights of the copyright holders.  Creative artists have sent letters to Congress and some have filed lawsuits to restrict the use of their works.  But if these subsequent works are not “speech” the AI developers may be exposed. Stay tuned, this may go all the way up.

Next:  A federal district court in D.C. just held that the product of AI is not capable of copyright protection because the statute contemplates human interaction.

If you have any questions or would like to discuss an issue involving AI and First Amendment protection, contact me at dan@dandowney.com.