You can't pick up a newspaper or magazine, watch a news program or social media post, or talk to a business owner without hearing about the latest rush to incorporate AI into all aspects of every business. While AI can be extremely helpful, bordering on miraculous, notwithstanding its occasional hallucinations, one may be surprised to learn, to their dismay, that the AI-generated work product and creations may not actually be solely owned by them or their business.
What most people, or at least many, don't understand or appreciate is that they don't fully own any of the materials created by AI, no matter how much time and money they put into its creation. When I use the term "own," I mean having the exclusive right to use and control the use of the end product, to be able to prevent others from using "your" AI-generated materials. If you can't prevent others from using it, you don't truly own it.
How did this come about? It starts with the concept that copyright, which is the principal source of protection and exclusive rights for literary works (books, magazines, manuals, brochures, etc.), advertisements, promotional material, computer programs, websites, social media postings, audio-visual works, film, music, art, photography, and the like, is not afforded to AI-generated works.
The U.S. Copyright Office has long held that copyrights are available only to works created by humans. AI-generated works are created by computers and algorithms, not humans. Therefore, the end product of AI-generated works is not protected by copyright and is free for anyone to copy and use as they wish. This position has been challenged in court, up to the Supreme Court, but to date it has been upheld by the courts. The courts have also long held that time and money spent ("sweat of the brow " is the legal term of art they use), no matter how extensive, do not get you copyright protection. The work must be sufficiently original and created by a human.
To date, prompts are not considered copyrightable either. Another fundamental copyright holding is that ideas in and of themselves are not protected. When you take an idea and a human fixes it into a tangible form, it becomes subject to copyright protection, but not the underlying idea. As a result, no matter how much time and money is expended, and how original the underlying idea is, if the end product is generated by prompts and an AI program, it is not protected by copyright.
All is not lost, however, if there is human input integrated into the AI-generated material. That input can be protected. When one creates an AI-generated work and then has human input added to it, or modifying it, those human contributions can be protected by copyright (if sufficiently original and substantial). Adding human contributions does not, however, create any copyright in the underlying AI material.
If one creates a hybrid work (AI- and human-generated) that can be registered with the U.S. Copyright Office and afforded some protection, one must still identify the AI- and human-generated portions and disclaim all of the AI-generated contributions from the overall registration.
Generally, you must have a copyright registration in hand to file suit to enforce your rights and to prevent others from using your materials. However, that hybrid registration will apply only to the non-AI/human-generated portion of the work. The more human-generated elements, the broader and stronger the suit will be. A mostly AI-generated work will have limited protection in any court of law.
Other forms of intellectual property might come into play, which afford some protection, such as rights of publicity, privacy, and trademark law, to protect some aspects of the AI-generated material. Even if they do, that doesn't protect the AI-generated parts of the works that are covered by copyright, patent, or trade secret law
It seems that a similar problem found in copyright law exists in patent law. In a patent application, one must list the actual inventor of the subject matter to be patented. At present, an AI program cannot be listed as an author on a U.S. Patent application. You have a similar situation where patents for AI-assisted inventions may qualify, but here again there must be significant human conception of the invention for the possibility of patentability to exist.
Next, we look to see if trade secret law would protect an AI-generated work. Trade secret law basically says that if you have some commercially valuable information, not generally available to the public, that you devised and that is kept under lock and key under appropriate state law guidelines, even if it's not copyrightable, it can still be protected from use by your competitors. For example, recipes are not protected under copyright law, but certainly the highly secret Coca-Cola recipe for its drinks is protected as a trade secret.
In order to be a trade secret, it must be a secret. Since AI is created in a third party's cloud, stored in that cloud, and accessible by the third party, and based on its terms and conditions, the third party determines how it's shared, if it's shared, and what parameters there are for its use, not the business that used prompts to create it. Additionally, if a business uploads trade secret information to an AI cloud service, are they risking losing the trade secret status on the uploaded material? What if a third party prompts the AI tool and now that it has your trade secrets in its data set, it uses that information to provide answers or solutions to a third-party prompt? What if a competitor actively seeks to gain access to your trade secrets by using specific prompts to find it?
Finally, a few states have passed or are contemplating laws that grant AI-generated materials copyright-like protection under state law, filling the gap created under copyright law for AI-generated materials. Those laws will certainly face a legal challenge and an uphill battle in being sustained, as U.S. law says that the U.S. federal copyright law preempts all state laws that deal with copyright and copyright-related matters.
AI is here to stay, and everyone will use it more and more, but beware that the gains it provides are not without some major IP-related risks.