According to CompTIA Community, one-third of all law firms are already using artificial intelligence in some way – including AI-generated material. Generative AI is proving particularly useful in marketing applications, and professionals in this field use AI-generated material for social media, websites, and various advertisements. However, Harvard Business Review points out that “generative AI has an intellectual property problem” – highlighting numerous challenges and unanswered questions. Companies implementing AI solutions may wish to explore some of these questions in more detail during a consultation with an experienced business law attorney in Louisiana. Consider the Business Law Group and call (504) 446-6506 to learn more about how you can ensure your legal rights are protected during a consultation.
What Is Generative AI?
Generative AI is a relatively new technology that can create various materials with basic human prompts. These AI-generated materials might be videos, images, or text. Some generative AI models can even create computer code, helping to build various apps and software solutions. While generative AI is not capable of “thinking” independently in the same way as a human, it can analyze vast amounts of data faster than any human mind.
Generally speaking, generative AI technology “crawls” databases (including the internet) to satisfy human prompts. For example, a user might ask a generative AI model to create an essay on pigeons. The AI model would then analyze web-based data on pigeons and generate original text built from a “mosaic” of various sources. The question, of course, is whether this “work” is truly original in the context of IP protection standards.
What Are IP Protection Standards?
IP protection standards represent a set of rules and requirements that uphold the rights of creators. Intellectual property may include virtually any creative “work,” including music, inventions, writing, visual designs, and so on. Within the larger category of intellectual property are more specific forms of protection, including copyrights, trademarks, and patents.
Various IP protection standards raise considerable questions regarding AI-generated material. First, there is the longstanding rule that copyrights only apply to works with human authors. Since AI-generated text and art have no human authors, it may fail to earn protection under existing copyright law. In addition, AI-generated art may violate existing intellectual property protections – including trademarks and copyrights. Finally, the “input” used to create AI-generated material might constitute additional IP violations. This is because some AI models allegedly analyze works illegally and without permission during the training process.
Can I Copyright AI-Generated Material?
Marketing professionals and business leaders may need to be careful when using AI-generated material in commerce, as these works may not gain copyright or trademark protection under current US law. In August of 2023, a federal judge in the United States denied an attempt to copyright an image generated by a so-called “neural network firm.” This judge pointed out that the CEO of this company listed the generative AI technology as the sole creator of the image. There was no human author, and therefore the image could not gain copyright protections. This ruling suggests that attempts to copyright or trademark AI-generated material (including designs and logos) will not be successful. However, some believe that the United States Patent and Trademark Office (USPTO) is willing to accept that their IP protection standards are somewhat antiquated – and these rules may change in the future to reflect the rise of AI.
Can I Train My AI Model on Copyrighted Materials?
Companies interested in creating new AI models inevitably need to “train” this software with a database of human-created content. Ideally, this database should be as extensive as possible. There is also evidence that training AI models on AI-generated material can “poison” or “cannibalize” the software – and this highlights the need to focus on human-created, original works.
However, many human creators have sued AI models for accessing their material without permission and without payment. For example, some AI companies faced accusations of accessing “darknet” databases of books and training their AI models on this content without purchasing the material first. The plaintiffs argue that their content was pirated, stolen, and used for profit. In the eyes of the authors, it is almost as if someone took their book from a store shelf, copied the general plotline, and then began selling an almost identical book.
The Fair Use Defense
Companies accused of accessing copyrighted materials to train AI models have generally responded with the fair use defense. “Fair use” is an exception to copyright protections, and it is one of the most important IP protection standards. Under this principle, creators can access copyrighted material and use it in certain ways. According to the Harvard Office of the General Counsel, fair use allows creators to “build upon” these materials and protects First Amendment rights. Examples of fair use include criticism, journalism, education, and research. Even parodies may constitute fair use. To utilize the fair use defense, those accused of copyright violations might need to demonstrate that they:
-
Did not profit in any way
-
Used the resulting work for educational purposes
-
Only used or copied a small portion of the work (such as a short quote)
The question is whether AI-generated material constitutes fair use. The USPTO specifically states that under its current IP protection standards, training an AI system is indeed fair use. While accessing copyrighted material to train an AI model is fair use, the legal status of AI-generated material is less clear. The “input” and “output” of these AI systems are not the same from a legal perspective – and this might be a topic worth discussing further alongside the Business Law Group.
Contact our Experienced Attorneys at Business Law Group Today
The potential of AI-generated material is considerable, but its applications also depend on the unique goals and priorities of each company. In the same way, the intellectual property issues associated with generative AI also depend on the specific factors surrounding each company. Many business leaders choose internet research as an initial first step into this complex field, but no amount of reading can actually build a comprehensive strategy to address IP protection standards. A more proactive approach might involve speaking with an experienced business law attorney in Louisiana. Consider the Business Law Group and call (504) 446-6506 today to continue this conversation in more detail.
Comments
There are no comments for this post. Be the first and Add your Comment below.
Leave a Comment