We asked the artificial intelligence chatbot ChatGPT the following question: “Are you infringing intellectual property with your work?” This was the answer we received:
“[T]he content I generate could potentially infringe on existing intellectual property rights, depending on the specific context and content of the input. It is the responsibility of the users of my responses to ensure that they are not violating any intellectual property laws or regulations.”
As we delve into this and similar questions, it’s important to consider the ripple effects creators and innovators will feel as AI technology quickly advances.
Increased public availability of AI has generated significant questions in the IP world [LINK TO APRIL BLOG]. We continue to keep a close eye on the development of artificial intelligence technologies as our clients continue to question how they can use artificial intelligence to increase their business’s efficiencies without infringing third-party IP and also how their creative work and intellectual property might be used and infringed upon by others using artificial intelligence technologies.
Can Artificial Intelligence Infringe My Intellectual Property?
As demonstrated by the response to our question that we received from ChatGPT, AI models willingly admit the potential for intellectual property infringement. However, we asked the same exact question to ChatGPT at another time, and we received a different response, this time claiming, “Any text that I generate is based on publicly available information and does not contain any proprietary or confidential information.” Although this answer addresses the potential use of confidential information, it does not address the potential for intellectual property infringement. In any event, this answer also suggests that these chats have not been programmed to give a canned response to questions concerning possible legal pitfalls.
Bard, Google’s AI system, responded by saying “I am not stealing intellectual property with my work. I am trained on a massive dataset of text and code, but I do not copy or plagiarize any of the content that I generate. I use my knowledge to create new and original content that is relevant to your requests.” Of course, this begs the question since it stated that it doesn’t copy or plagiarize content that it generates, but it doesn’t address third-party content that it might use to generate its own content.
These chats show inconsistencies in what they, themselves, are saying about the use of third-party intellectual property in their generation of content for their users. There are additional concerns when we consider programs that generate “AI art.”
Repeated questions about whether these programs steal art along with numerous lawsuits being filed over the use of human-created art in AI programs have put a magnifying glass on these programs. As an IP attorney, the evidence of third-party works being used by AI is relatively insurmountable at this point. Here is one example of a graphic designer posting on Twitter about original work being used by AI to create additional content. Existing works are being “fed” into these programs as part of the data set which is then used as the basis to generate the images requested by users.
Although numerous lawsuits have been filed, few have resulted in law that can be considered or cited. Until the courts (or Congress) provide guidance on these issues, it remains to be seen how intellectual property laws might be applied in the context of content generated by AI technologies.
Who Owns What AI Produces?
Questions also remain about who actually owns the content that the AI programs are producing. Is it the user who requested the content or prompted the response? Is it the AI itself? Or the humans that created the AI or provided the AI with the data to be able to create the content or response?
Currently, the intellectual property laws hold that only humans have a right to creative authorship. This means an AI model cannot claim ownership over the work it produces, even if it can closely mimic the human creative process.
The potential answers lie somewhere between (1) all work generated by AI being considered public domain and (2) a debate over whether the user of the AI or the inventor of the AI owns it. After all, the inventor did not submit the entry that generated the work, but the user did not create the program that generated the response. Or should our laws be amended to allow the AI model itself to claim ownership?
It’s likely the courts (and Congress) will have extensive work cut out for them to clarify these questions along with who will be held responsible should AI be found to have violated someone else’s intellectual property rights. In the meantime, the terms and conditions of a given AI program may give further insight into how the creators of the AI intended for IP rights to be held.
At McDermott IP Law, we believe in innovation, and we assist our clients to protect and enforce their innovation. Artificial intelligence will require the balancing of the innovation of the AI itself and human innovation that might be used by such AI programs. If you have an intellectual property question related to artificial intelligence or need help protecting your innovation, contact McDermott IP Law.
McDermott IP Law
Latest posts by McDermott IP Law (see all)
- From Creation to Commercialization: The Lifecycle of IP Assets - April 15, 2024
