It Wasn’t Me, It Was the AI: Intellectual Property and Data Privacy Concerns With Nonprofits’ Use of Artificial Intelligence Systems
In today's rapidly changing technological landscape, artificial intelligence (AI) is making headlines and being discussed constantly. To be sure, AI provides a powerful tool to nonprofits in creating content and exploiting for countless cost-effective purposes. As nonprofit executives, you may be wondering how AI intersects with intellectual property and data privacy law, and how it could affect your organization. While the full extent of the implications will only be fully understood after some history with the use of AI, some of the issues are already predictable.
One of the main concerns when it comes to AI and intellectual property is how copyright laws impact the use of AI. In the copyright realm, there are two prominent issues—indeed, two sides of the copyright coin—for nonprofits to consider:
Ownership of AI-created content: If your nonprofit uses AI to create content, you need to clarify who owns that content; it will be difficult to protect that content against use by third parties. That is because, under United States law (and the law of nearly all other jurisdictions), copyright is only recognized in works created by human beings. AI-generated content is not eligible for copyright protection. As such, one would not be able to protect against third parties who copy such content and use it for their own purposes.
The best defense against such taking, then, would be to add human creation into the AI-generated work. While any copyright protection would only vest in those human-authored additions/revisions, to the extent a third party’s copying included those changes, it could support a claim of copyright infringement, and would give the nonprofit a basis to stop that third party’s use of the work. Note, though, that a copy that did not incorporate those human-authored aspects would not violate the rights of the authoring entity.
Copyright infringement: AI's ability to make copies of data, including copyrighted material, raises concerns about potential copyright infringement. For example, AI large language models learn by analyzing vast amounts of data from the materials it is provided, which could include the entirety of internet content accessible at the time the AI system was trained. Obviously this could, and necessarily would, include copyrighted material. In undertaking such training, the entity training the AI system necessarily copies those materials, though the copying is not long-term and the copies are discarded. While fair use may be a defense to that kind of copying, it is a complex legal issue, and you as the trainer of the AI system may need to be prepared for potential litigation.
That said, the issues with training copies are more of a concern for the owners of the AI system, and if the system you are using is one provided by a third party, copyright infringement arising from training may not be your concern.. But there is another consideration. We are starting to see issues where AI systems are “creating” content that seems suspiciously similar to existing content. To the extent your nonprofit uses an AI system—even one provided by a third party for your use—to create content (text, images, videos, etc.) that you post on your site or otherwise distribute, where that content actually copies protected third-party works, your nonprofit could be liable for copyright infringement.
Here, it is important to look at the agreements governing your use of the system. Just as you would expect an artist you hire to stand behind their work, you could try to have the owner of the AI system provide warranties against infringing creations. It is certainly unlikely that off-the-shelf products are going to include such protections in their user agreements. But as AI business develops, there will likely be bespoke work done for clients, and in those relationships such contractual protection could be available. And even assuming the provider offers reps and warranties, and even indemnification, against such copyright infringement claims, that may not be enough. A claim for copyright infringement against materials you post on your site will be brought against you. Even if the AI-provider is ultimately responsible under the agreement, if it does not have the resources to defend you (if, for example, it is out of business), you won’t be off the hook. With all of this in mind, it is imperative to carefully consider the product and the provider before using AI to create content that you plan to disseminate through your website or otherwise.
Right of Publicity and Privacy Concerns
Another area in which the use of AI could present issues is that of the rights of publicity and privacy. The right of publicity protects individuals' rights to control their name, image, and other identifiers for commercial purposes. One could imagine a scenario where AI-created content includes such materials without the authorization of the individual identified. Under state right of publicity laws, courts have commonly found liability for the use of look-alikes (and even sound-alikes) even where the actual person’s image (or voice) is not used.
Additionally, AI systems could collect personal information of individuals and lead to the company employing such systems storing and using such personal information without authorization from such individuals. Indeed, considering that the system may have had access to a broad universe of data, it is likely such data included individuals’ personal information. Any use of an AI system should be carefully tailored to protect against unintended use and collection of third-party data.
It is also possible for an AI system to re-identify personal information from data that was previously de-identified (as a matter of course or upon a request of an individual). In that case the company could be in violation of its own consumer-facing privacy statement, as well as, potentially, the law.
Finally, AI systems may scrape personal data to use in connection with training, and then such information may be used otherwise without authorization.
To safeguard your nonprofit's reputation and minimize legal pitfalls, consider the following:
- Carefully review content for potential issues with the identification of individuals;
- Implement gatekeeping measures to protect sensitive donor and employee data. Techniques like CAPTCHA can help keep AI from scraping information from your website.
Of course, these are just a few of the issues we expect to arise in IP and data privacy from the use of AI systems. What is clear, though, is that it is essential that nonprofits are vigilant as to how AI impacts your organization's intellectual property and privacy practices. While AI offers exciting possibilities for nonprofits, it also poses legal challenges that require careful consideration. Stay informed, work with legal and IT professionals, and update your policies to ensure ethical AI usage and protect your nonprofit's reputation and avoid legal issues. By understanding the implications of AI and adopting responsible practices, your nonprofit can harness the power of technology while safeguarding your stakeholders' interests.
"It Wasn’t Me, It Was the AI: Intellectual Property and Data Privacy Concerns With Nonprofits’ Use of Artificial Intelligence System,” published in Board & Administrator, December 2023, Vol. 40 , No. 4, Copyright 2023.