The Best Way to Protect Your AI-Generated Intellectual Property? Shhh…It’s a Secret
December 12, 2023
Artificial intelligence has permeated every industry that involves even the most minimal use of computers—i.e., every industry on Earth. And as the AI wave has swept the commercial world, it has left in its wake several important questions about the bounds of intellectual property law, some quite philosophical and others much more pragmatic: Can AI create art? Can AI invent something? Who owns what AI makes? Can AI platforms themselves be protectable intellectual property?
Given that commercial AI tools are routinely described as “cutting-edge,” it is no surprise that legislatures and courts have yet to answer many of these questions in meaningful ways. Nonetheless, for a company using AI or considering using AI, it is important to understand where intellectual property pitfalls may lie.
What is AI?
In the simplest terms, artificial intelligence generally refers to the ability of a computer or computer-controlled robot to perform tasks typically associated with intelligent beings.1 AI is already an integral part of daily life—in home digital assistants, social media algorithms, targeted advertising, healthcare, manufacturing, banking, and more.2 In fact, “AI” is so in vogue, many companies are rebranding themselves as AI companies, despite the fact that many of their products have not changed in years.3
“Generative AI” refers more specifically to artificial intelligence systems that are designed to create new and original content based on the data they are trained on (ChatGPT and DALL-E are examples).4 Generative AI utilizes large data sets in a process called “deep learning” to analyze common patterns and arrangements to “generate” (hence the name) novel outputs.5 For example, a generative AI model trained on massive amounts of fiction writing could, by learning to recognize setting, characters, plot, themes, etc., write new stories based on its learned framework. The more data an AI model accesses, the more sophisticated these outcomes are likely to be.6 Because the outputs of generative AI are novel, rather than reproductions or collages of work that already exists, they present especially interesting questions in the field of intellectual property.
Current Legal Landscape
Even though AI’s influence has grown exponentially, there are very few laws or regulations in place to address the issues raised by AI.7 As a result, court systems will be left to make sense of the issues until the law catches up.
And though there have not yet been many intellectual property cases involving AI, we are hearing rumblings of cases that will soon seize the world’s attention.8
No matter the outcome of these cases, companies—especially global entities—will certainly need to be more careful than ever to (1) keep their own data protected, (2) avoid liability for infringement claims, and (3) keep up with changing laws and regulations in various jurisdictions.
Protecting AI Platforms Themselves
One immediate concern, given the increasing prevalence of AI, is how technology companies and AI developers might protect their own AI platforms and tools. A possible approach—keep it secret. Practitioners hoping to patent AI systems face an uphill battle anyway, since current patent jurisprudence is unfavorable for software in general and will likely be even less favorable for AI. On the other hand, trade secret law may offer an alternative form of protection.
Though patents offer some of the most concrete protections for intellectual property, patenting AI systems is often difficult because AI claims run into many of the same §101 pitfalls as other algorithms. It is impermissible to patent an abstract idea, and, under the Supreme Court’s Alice framework, many software-related inventions have been deemed abstract ideas, and thus unpatentable.9
AI is particularly vulnerable under the Alice framework because it often acts as a black box—it’s not always clear what algorithm a claim actually utilizes or what process that algorithm is performing. A vague description of an algorithm’s function is not enough to escape classification as an abstract idea.10
And even if an inventor provides a succinct but thorough description, AI claims still fall when they fail to add any inventive step.11 Under Alice, AI must do more than merely complete tasks more efficiently than a human.12
In contrast, trade secret law is actually surprisingly well-adapted to protect a host of AI-related material—including training data, AI software code, input parameters, and AI-generated output that is intended only for internal and confidential use.13 The federal Defend Trade Secrets Act protects “all forms and types of financial, business, scientific, technical, economic, or engineering information” where (1) the owner has taken reasonable measures to keep such information secret and (2) the information derives independent economic value, actual or potential, from not being generally known or readily ascertainable by others.14 Although trade secret protection requires more maintenance than patent protection—trade secret holders need to continuously work to protect their secrets—it could more easily cover AI-related material. Also, unlike patents, which have a limited lifespan, trade secrets can theoretically live forever.15
Of course, once the AI output is public it cannot be protected as a trade secret—something that is public cannot also be secret.
Protecting the Outputs of AI Platforms
Protecting AI-generated outputs is even more challenging than protecting AI systems themselves. Global legal systems have begun grappling with the question of whether to grant IP protection to works created by AI. At this point, a few people have filed test cases to see how courts will address purported AI-artists and AI-inventors.16 Generally, though not uniformly, courts have deemed creative works and inventions created by AI unprotectable.
Notably, Stephen Thaler has filed several cases in various jurisdictions with the ultimate goal of granting patents to his AI platform (DABUS).17 Thaler claims DABUS was the sole inventor of “Neural Flame,” a “light beacon that flashes in a new and inventive manner to attract attention,” and “Fractal Container,” a “beverage container based on fractal geometry.”18 And though the Federal Circuit ruled that under U.S. law “there is no ambiguity: the Patent Act requires that inventors must be natural persons; that is, human beings,”19 not every country has taken the same stance. For example, South Africa granted a patent to Thaler’s DABUS.20 A lower court in Australia did the same, though it was later overturned.21
Copyright offices and courts have also generally denied copyright protection to AI-created works. Thaler’s other AI system, the “Creativity Machine,” allegedly created a piece of art called “A Recent Entrance to Paradise,” and that was denied a copyright by the Copyright Office.22 Upholding the decision, a judge in the District of Columbia held that “United States copyright law protects only works of human creation.”23 In cases where humans and AI have combined to create a work, the Copyright Office has given protection to the human-created portions and denied protection to the AI-created portions.24 So, if you are seeking to use copyright law to protect the output, it is important to keep a good record of which elements were human-created.
But again, trade secret law to the rescue. Unlike the ownership requirements in copyright and patent law, the ownership requirements for a trade secret depend upon secrecy rather than the circumstances of the trade secret’s creation. In plain language, trade secrets do not need to be created by a human being.25 So, AI-created works that are not public could qualify as protectible trade secrets if the other elements of a trade secret are also met. Again, this could only cover outputs that are kept secret, and would not cover artworks like “A Recent Entrant into Paradise” which are available for all the public to see.
Takeaways
This is just a sketch of the conflicts between AI and intellectual property. Practitioners working in the field should operate carefully and keep an eye on the evolving law. But for now, protecting both AI-generated works and AI platforms themselves may be most safely done via trade secret law. For other IP protections, such as patent and copyright, it will require a more thorough analysis and good documentation of the elements that were created by humans.
The AI landscape is rapidly changing and there is no clear, agreed-upon way of handling AI and AI-created works—either globally or in the United States,26 though the EU is currently working on the Artificial Intelligence Act, which will be the world’s first rules on AI.27 As AI becomes ubiquitous, the gap between old law and new technology will become wider, and lawmakers, courts, and practitioners will need to do some work to bring them together.
1 Artificial Intelligence (AI), U.S. DEPT. OF STATE, www.state.gov, https://www.state.gov/artificial-intelligence/ (Accessed Oct. 26, 2023); What is artificial intelligence (AI)?, IBM, www.ibm.com, https://www.ibm.com/topics/artificial-intelligence (Accessed Oct. 26, 2023)
2 Artificial Intelligence: What it is and why it matters, SAS, www.sas.com, https://www.sas.com/en_us/insights/analytics/what-is-artificial-intelligence.html#:~:text=Artificial%20intelligence%20(AI)%20makes%20it,learning%20and%20natural%20language%20processing (Accessed Oct. 26, 2023).
3 You Can’t Trust the AI Hype, PALLADIUM MAGAZINE, https://www.palladiummag.com/2023/08/03/you-cant-trust-the-ai-hype/ (Accessed Oct. 26, 2023).
4 What is generative AI?, MCKINSEY & CO., www.mckinsey.com, https://www.mckinsey.com/featured-insights/mckinsey-explainers/what-is-generative-ai (Accessed Oct. 26, 2023); Introducing ChatGPT, OPENAI, openai.com, https://openai.com/blog/chatgpt (Accessed Oct. 26, 2023)
5 Owen Hughes, Generative AI Defined: How it Works, Benefits, and Dangers, TECHREPUBLIC, https://www.techrepublic.com/article/what-is-generative-ai/ (Accessed Oct. 27, 2023).
6 Id.
7 See, e.g., Artificial Intelligence Legislation Tracker, BRENNAN CENTER, https://www.brennancenter.org/our-work/research-reports/artificial-intelligence-legislation-tracker (Accessed Oct. 27, 2023); Global AI Legislation Tracker, IAPP, https://iapp.org/resources/article/global-ai-legislation-tracker/ (Accessed Oct. 27, 2023).
8 See, e.g., Generative Artificial Intelligence: Federal Litigation Tracker, WESTLAW, https://1.next.westlaw.com/Document/I64d1932b1f2911ee8921fbef1a541940/View/FullText.html?transitionType=Default&contextData=(sc.Default)&isplcus=true&firstPage=true (Accessed Oct. 27, 2023).
9 See, e.g., Angel Technologies Group LLC v. Facebook Inc., 2022 WL 3093232 (C.D. Cal. Jun. 30, 2022); Blue Spike, LLC v. Google Inc., 2015 WL 5260506 (N.D. Cal. Sept. 8, 2015).
10 See, e.g., eResearchTechnology, Inc. v. CRF Health, 186 F.Supp. 3d. 463 (W.D. Pa. May 10, 2016); FacetoFace Biometrics, Inc. v. Apple, Inc., 2023 WL 2561758 (E.D. Mo. Mar. 17, 2023).
11 See e.g., Health Discovery Corporation v. Intel Corporation, 577 F.Supp.3d 570 (W.D. Tex. Dec. 27, 2021).
12See, e.g., Hyper Search, LLC v. Facebook, Inc., 2018 WL 6617143 (D. Del. Dec. 17, 2018).
13 See generally, Defend Trade Secrets Act, 18 U.S.C. § 1836.
14 18 U.S.C. § 1839(3).
15 Frequently Asked Questions: Trade Secrets, WIPO, https://www.wipo.int/tradesecrets/en/tradesecrets_faqs.html (Accessed Oct. 27, 2023).
16 See The Artificial Inventor Project, https://artificialinventor.com/ (Accessed Oct. 27, 2023).
17 See, e.g., Thaler v Commissioner of Patents [2021] FCA 879 (Jul.30, 2021) (Austl.).
18 U.S. Patent Application No. 16/524,350; U.S. Patent Application No. 16/524,532.
19 Thaler v. Vidal, 43 F.4th 1207, 1210 (Fed. Cir. 2022); see also Thaler v. Vidal, 143 S.Ct. 1783 (2023) (denying writ of certiorari).
20 See, e.g., DABUS Gets Its First Patent in South Africa Under Formalities Examination, IPWatchdog, https://ipwatchdog.com/2021/07/29/dabus-gets-first-patent-south-africa-formalities-examination/id=136116/ (Accessed Oct. 27, 2023); Acceptance of Complete Specification, https://ipwatchdog.com/wp-content/uploads/2021/07/AP7471ZA00-Notice-of-Acceptance-1.pdf (Accessed Oct. 27, 2023).
21 See In the Courts: Australian Court finds AI systems can be “inventors,” WIPO, https://www.wipo.int/wipo_magazine/en/2021/03/article_0006.html#:~:text=Financial%20Reporting%20Oversight-,In%20the%20Courts%3A%20Australian%20Court,AI%20systems%20can%20be%20%E2%80%9Cinventors%E2%80%9D&text=In%20a%20world%20first%2C%20a,of%20the%20Australian%20patent%20regime. (Accessed Oct. 27, 2023); Australian Court Overturns AI Inventorship Ruling, JD Supra, https://www.jdsupra.com/legalnews/australian-court-overturns-ai-1320594/ (Accessed Oct. 27, 2023).
22 See Thaler v. Perlmutter, C.A. No. 22-1564 (BAH), 2023 WL 5333236, at *1 (D. D.C. Aug. 18, 2023).
23 Id. at *10.
24 In re Zarya of the Dawn, U.S. COPYRIGHT OFFICE, https://www.copyright.gov/docs/zarya-of-the-dawn.pdf (Accessed Oct. 27, 2023).
25 See, e.g., 18 U.S.C. §1839(4) (defining trade secret owner).
26 As discussed above, there are conflicts in global policy and even between courts in the same jurisdiction.
27 See EU AI Act: first regulation on artificial intelligence, E.U. PARLIAMENT, https://www.europarl.europa.eu/news/en/headlines/society/20230601STO93804/eu-ai-act-first-regulation-on-artificial-intelligence (Accessed Oct. 27, 2023).
This memorandum is a summary for general information and discussion only and may be considered an advertisement for certain purposes. It is not a full analysis of the matters presented, may not be relied upon as legal advice, and does not purport to represent the views of our clients or the Firm. Scott Pink, an O’Melveny special counsel licensed to practice law in California, Marc J. Pensabene, an O'Melveny partner licensed to practice law in New York, Coke Morgan Stewart, an O'Melveny senior counsel licensed to practice law in the District of Columbia and Virginia, Paige Hardy, an O’Melveny associate licensed to practice law in California, and Grace McFee, an O’Melveny associate licensed to practice law in California, contributed to the content of this newsletter. The views expressed in this newsletter are the views of the authors except as otherwise noted.
© 2023 O’Melveny & Myers LLP. All Rights Reserved. Portions of this communication may contain attorney advertising. Prior results do not guarantee a similar outcome. Please direct all inquiries regarding New York’s Rules of Professional Conduct to O’Melveny & Myers LLP, Times Square Tower, 7 Times Square, New York, NY, 10036, T: +1 212 326 2000.