Introduction
The rapid development of artificial intelligence ("AI") is revolutionizing industries and will reshape the global economy in the coming years. As U.S. companies invest heavily in AI research and development, seeking to bring groundbreaking products to market, they will undoubtedly need to protect their intellectual property ("IP") rights against foreign competitors.
The International Trade Commission ("ITC"), with its powerful tools for combating infringing imports under Section 337 of the Tariff Act of 1930, will likely see an increasing caseload involving AI products. However, the intangible nature of AI technology, which often involves software and training data rather than physical products, may test the limits of the ITC's jurisdiction.
Recent court decisions, such as the Federal Circuit's ruling in ClearCorrect, may narrow the types of AI products that can be litigated at the ITC and litigators may need to adapt their approach to effectively address the coming wave of AI-related disputes.
This article explores the complex legal and strategic landscape surrounding AI and the ITC. It begins by providing background on the ITC's jurisdiction under Section 337 and its powerful remedies. Next, it examines the ongoing AI revolution and its potential economic impact.
The article then analyzes how Federal Circuit case law may affect the ITC's jurisdiction over four key areas of AI technology: training data, hardware, software, and AI-first products. Finally, it posits litigation strategies for the types of AI disputes likely to arise at the ITC in the coming years.
Background
The U.S. International Trade Commission's Intellectual Property Jurisdiction and Its Powerful Remedies
The ITC is a quasi-judicial federal agency that plays a crucial role in protecting domestic industries from unfair trade practices. Among its various responsibilities, the ITC has the authority to investigate and adjudicate complaints alleging infringement of IP rights under Section 337 of the Tariff Act of 1930, codified as 19 U.S.C. § 1337.
Under Section 337, participants in U.S. "domestic industries" may seek relief against infringing imports in the form of exclusion orders and cease and desist orders. Exclusion orders, enforced by U.S. Customs and Border Protection, bar the importation of infringing articles into the United States. Cease and desist orders prohibit named respondents from distributing infringing articles already imported into the country.
Section 337 grants the ITC broad jurisdiction over unfair acts including the importation, sale for importation, or sale after importation of "articles" that infringe a U.S. patent, trademark, or copyright. This generally broad jurisdiction enables the ITC to investigate and remedy infringement by foreign manufacturers and importers, even when they may not be subject to the jurisdiction of Article III courts.
The Artificial Intelligence Revolution
The age of AI has arrived, and it is transforming our world at a record pace. With rapid advancements in machine learning, natural language processing, and computer vision, AI systems are starting to handle complex tasks previously thought to be uniquely "human," such as image recognition, visual reasoning, and language comprehension.
While AI is still in its infancy, if effectively developed and deployed, its economic impact could be immense.1 McKinsey & Company, for example, predicts that AI could automate work activities that take up between 60 and 70% of workers' time today.2 AI is estimated to enable labor productivity growth of about 0.1 to 3.4% annually through 2040, depending on the speed and extent of adoption.3
Due to the resulting combination of skyrocketing labor productivity and the creation of new, high-value added industries, AI has the potential to add between $2.6 and $4.4 trillion dollars annually to the global economy.4 Funding for AI, then, is predictably on fire: $12 billion was funneled into AI investment in just the first five months of 2023.5
Investments in AI technology have grown 74% annually between 2017 and 2022, resulting in vigorous development in four key frontiers: AI training data and algorithms, computer hardware to run AI systems, software utilizing AI, and AI-first products.6
Waves of Articles Challenged at the ITC
Previous Generations of Challenged Articles
Over the past few decades, the types of products litigated at the ITC have evolved in tandem with changing economic landscapes and technological advancements. In the late 1980s, cases focused on antibiotics, batteries, printers, memory chips, textiles, and manufacturing tools.7 Moving into the 1990s, semiconductors, hard disk drives, fax machines, satellite TV, CDs, HIV screening kits, acid-washed denim garments, and chemotherapy drugs came to the forefront.8
As we entered the new millennium, the ITC saw an influx of cases related to flash memory, network communications, rewritable CDs, LCD monitors, cameras, cell phones, and semiconductor manufacturing tools and methods.9 The latter half of the 2000s brought litigation concerning LCD displays, TVs, laser diodes, GPS devices, and telecommunications.10
In the 2010s, the ITC saw an era dominated by smart phones, network devices, small consumer electronics, video game consoles, pharmaceuticals, and medical devices.11
Most recently,12 from 2015 to 2020, the ITC has grappled with cases involving telecommunications, consumer electronics, automotive, transportation, and manufacturing industries, showcasing the breadth and complexity of modern trade disputes.13
The Next Wave of AI Software and Products
AI systems are dependent on their vast sets of training data, and the source code and machine learning algorithms that make sense of it all. The training data includes "unlabeled data" — text, images, audio, and video taken directly from sources including the internet — and "labeled data" — human-tagged data specifically for AI training (for example, tagging spam in an email inbox or tagging the pictures with a motorcycle in a reCAPTCHA test).14
Specialized hardware is needed to process their datasets and algorithms for AI systems. These include high-performance CPUs to manage the overall workload, GPUs to accelerate computationally intensive tasks like training neural networks, high-bandwidth memory to quickly feed data to the processors, fast storage to hold the training data, and high-speed networking to connect multi-GPU setups.
The technology's rapid progress has and will continue to lead to a new wave of AI-first products, including, for example, industrial robots using computer vision and machine learning algorithms to assist with assembly, manufacturing, quality control, inspection, and packaging. Modern smartphones having built-in AI for voice control, biometric authentication, enhanced photography, and augmented reality. And autonomous vehicles using a suite of sensors and AI algorithms to perceive their environment and make real-time driving decisions.
AI is also enabling innovative software solutions across various industries. Healthcare applications assist doctors with analyzing medical imaging, discovering new drugs, and tailoring treatments to patients.
Financial institutions apply machine learning to detect fraud, assess risk, and automate trading. Manufacturers deploy computer vision systems to identify defects, predict equipment failures, and optimize product designs. And cybersecurity companies use deep learning to detect new malware and identify network intrusions. These AI-driven software products can improve efficiency, reduce costs, and achieve better results for organizations.
The United States currently leads in the production of state-of-the-art AI models and products, with class-leading models being developed by U.S.-based powerhouses Alphabet (Google), Amazon, Apple, IBM, Microsoft, Meta (Facebook), NVIDIA, and OpenAI.15 This area will undoubtably be a fiercely competitive global race for years and decades to come.16
AI Software and Product Disputes at the ITC
Digital vs. Physical Imports Under Section 337
In a hotly contested appeal from the ITC,17 the Federal Circuit held that the agency does not have jurisdiction over digital imports because they are not "articles" under Section 337.18
The court analyzed the Commission's decision under the Chevron19 framework and held that in view of the text, statutory context, and dictionaries from the passing of the Tariff Act of 1930, "articles" are limited to "material things." While the decision in ClearCorrect was in the context of patents, Section 337′s "article" requirement also applies to design patents, copyrights, and trademarks.20
The accused articles in ClearCorrect were digital 3D models of patients' teeth, which were designed in Pakistan and sent to the U.S. for ClearCorrect to manufacture plastic teeth aligners.21
In other words, no physical objects were being shipped to the U.S. The ALJ at the ITC determined, and the Commission agreed, that these digital 3D models were "articles" under Section 337 and thus that the ITC had jurisdiction.22 The ITC then issued a cease-and-desist order against ClearCorrect.23
On appeal, the Federal Circuit determined that under Chevron step-one, "articles," as used in Section 337, unambiguously means "material things" and "excludes purely digital data."24 In so finding, the court referenced nine contemporaneous and modern dictionary definitions of an "article."25
The court determined that under Chevron step-two, even if the statute was ambiguous, the ITC's broader interpretation of "articles" was unreasonable in view of the plain meaning, legislative history, statutory context, and practicalities of stopping electronic transmissions at the border.26
In her concurrence, Judge O'Malley invoked the major questions doctrine, which removes statutory interpretation inquiries from the Chevron framework in cases that are so "extraordinary" that Congress likely did not intend to delegate interpretative authority.27
In dissent, Judge Newman argued the majority's limitation of "articles" to "tangible" goods was contrary to Section 337 and precedent.28 She argued that Section 337 was meant to broadly protect U.S. industries, including from infringing digital goods.29
According to Judge Newman, removing the "preeminent form of today's technology" from the ITC's jurisdiction, locks the agency into "technological antiquity."30 Judge Newman penned a similarly vigorous, 21-page dissent from the court's denial of an en banc rehearing.31
The ITC's Jurisdiction Over Four Key Areas of AI
AI Training Data
AI training data, while taking immense resources to develop and organize, may not qualify for U.S. copyright protections.
To be eligible for U.S. copyright protection, (1) training data must fall within the statutory definition of a protectable "compilation,"32 (2) the author must "make the selection or arrangement independently,"33 and (3) the compilation of data must possess "at least some minimal degree of creativity."34
However, even if training data does qualify for U.S. patent or copyright protections, the data is not a "tangible" article and therefore would not be within the ITC's jurisdiction under ClearCorrect.35
Computer Hardware for AI Systems
Computer hardware for AI systems can be considered an extension of the computer chips that have routinely been litigated at the ITC. Computer hardware including CPUs, GPUs, networking chips, and memory chips are aggressively protected by U.S. patents and will likely be among the next generation of commonly litigated articles at the ITC.
Software That Leverages AI
Software that uses AI to accomplish computer-based tasks is routinely protected by U.S. copyrights and sometimes by U.S. patents. However, so long as the software is not tied to a "tangible" article, for example being loaded onto and sold via a CD or memory stick, AI software alone may fall outside the jurisdiction of the ITC under ClearCorrect.36
However, there may be a creative way for AI software patent holders to pursue infringement at the ITC even after ClearCorrect: indirect infringement. The Federal Circuit held in Suprema that the ITC properly found induced infringement of a U.S. software patent after importation into the United States.37
In that case, Suprema made fingerprint scanner devices that required connection to a computer running a certain companion software.38 The fingerprint scanners were shipped to the United States complete with the tools and instructions for the customer to develop the companion software, called a "software development kit" or "SDK."39
Suprema sold their fingerprint scanner to U.S.-based Mentalix, Inc. and collaborated with Mentalix to develop the companion software that practiced a claim of a U.S. software patent.
This collaboration, the Commission concluded, constituted inducement by Suprema.40 In summary, a hardware importer cannot insulate itself from the ITC's jurisdiction by shipping hardware stripped of its software and actively encouraging its customers to subsequently develop or download the infringing software upon the article's arrival in the U.S.
Consider, for example, a hypothetical scenario where medical imaging equipment is made abroad and its manufacturer boasts of the ability to leverage certain AI-based software to enhance imaging or conduct pre-readings.
Assume that the manufacturer works with its U.S. customers to
develop and install the companion AI-based software after
importation, rather than shipping the equipment with the AI
software pre-loaded.
The owner of a U.S. patent covering that AI-based software may be
able to allege induced infringement against the manufacturer based
on induced infringement; a potential narrow workaround to the
holding in ClearCorrect.
AI-First Products
AI-first products are the hottest products on the market and will undoubtably be the subject of extensive ITC litigation for years to come. These products may be litigated as practicing product patents or as "tangible" vehicles for AI software-related patents and copyrights.
Take, for example, the wildly anticipated "Rabbit R1," a portable device that is billed as a revolutionary "consumer AI device" running "rabbit OS," a proprietary AI model that "can infer and model human actions on computer interfaces by learning users' intention and behavior."41 Given the hype surrounding AI-first products like the Rabbit R1,42 these types of products will likely be a recurring theme in ITC litigation going forward.
Conclusion
As AI reshapes the global economy, the ITC will likely see an influx of AI-related litigation. While the agency's jurisdiction over digital goods remains limited after ClearCorrect, the ITC will remain an attractive venue for stakeholders to safeguard their AI-related IP due to its potent remedies and expertise in IP disputes.
Companies and litigators must remain vigilant in monitoring the evolving legal terrain at the ITC and develop proactive strategies to assert their IP rights and maintain their competitive edge in this new era of technological advancement.
Footnotes
1. See Yiwen Lu, Generative A.I. Can Add $4.4 Trillion in Value to Global Economy, Study Says, N.Y. TIMES at 2–3 (June 14, 2023).
2. The Economic Potential of Generative AI, McKinsey & Co. (June 2023). Michael Chui, et al., The Economic Potential of Generative AI, McKinsey & Co. at 3 (June 2023).
3. Id. at 44–46.
4. Id. at 9–11, 24–26.
5. Id. at 5.
6. Id. at 5. The authors define "AI-first products" as products that are that are fundamentally built around AI technologies.
7. Annual Report, Fiscal Year 1985, U.S. Int'l Trade Comm'n.; Annual Report, Fiscal Year 1986, U.S. Int'l Trade Comm'n.; Annual Report, Fiscal Year 1987, U.S. Int'l Trade Comm'n.; Annual Report, Fiscal Year 1988, U.S. Int'l Trade Comm'n.; Annual Report, Fiscal Year 1989, U.S. Int'l Trade Comm'n.
8. Annual Report, Fiscal Year 1990, U.S. Int'l Trade Comm'n.; Annual Report, Fiscal Year 1991, U.S. Int'l Trade Comm'n.; Annual Report, Fiscal Year 1992, U.S. Int'l Trade Comm'n.; Annual Report, Fiscal Year 1993, U.S. Int'l Trade Comm'n.; Annual Report, Fiscal Year 1994, U.S. Int'l Trade Comm'n.; Annual Report, Fiscal Year 1995, U.S. Int'l Trade Comm'n.; Annual Report, Fiscal Year 1996, U.S. Int'l Trade Comm'n.; Annual Report, Fiscal Year 1997, U.S. Int'l Trade Comm'n.; Annual Report, Fiscal Year 1998, U.S. Int'l Trade Comm'n.; Annual Report, Fiscal Year 1999, U.S. Int'l Trade Comm'n.
9. Year in Review, Fiscal Year 2000, U.S. Int'l Trade Comm'n.; Year in Review, Fiscal Year 2001, U.S. Int'l Trade Comm'n.; Year in Review, Fiscal Year 2002, U.S. Int'l Trade Comm'n.; Year in Review, Fiscal Year 2003, U.S. Int'l Trade Comm'n.; Year in Review, Fiscal Year 2004, U.S. Int'l Trade Comm'n.
10. Year in Review, Fiscal Year 2005, U.S. Int'l Trade Comm'n.; Year in Review, Fiscal Year 2006, U.S. Int'l Trade Comm'n.; Year in Review, Fiscal Year 2007, U.S. Int'l Trade Comm'n.; Year in Review, Fiscal Year 2008, U.S. Int'l Trade Comm'n.; Year in Review, Fiscal Year 2009, U.S. Int'l Trade Comm'n.
11. Year in Review, Fiscal Year 2010, U.S. Int'l Trade Comm'n.; FY 2011 at a Glance, Intellectual Property Import Investigations, U.S. Int'l Trade Comm'n.; FY 2012 at a Glance, Intellectual Property Import Investigations, U.S. Int'l Trade Comm'n.; FY 2013 at a Glance, Intellectual Property Import Investigations, U.S. Int'l Trade Comm'n.; FY 2014 at a Glance, Intellectual Property Import Investigations, U.S. Int'l Trade Comm'n.
12. The ITC has not published an Annual Report since its fiscal year 2020.
13. See Brian Busey & Daniel Muino, What Case Trends Reveal About Life Sciences Results At ITC, LAW360 (Oct. 16, 2023, 5:26 PM EDT); FY 2015 at a Glance, Intellectual Property Import Investigations, U.S. Int'l Trade Comm'n.; FY 2016 at a Glance, Intellectual Property Import Investigations, U.S. Int'l Trade Comm'n.; FY 2020 at a Glance, Intellectual Property Import Investigations, U.S. Int'l Trade Comm'n.; FY 2017 at a Glance, Intellectual Property Import Investigations, U.S. Int'l Trade Comm'n.; FY 2018 at a Glance, Intellectual Property Import Investigations, U.S. Int'l Trade Comm'n.; FY 2019 at a Glance, Intellectual Property Import Investigations, U.S. Int'l Trade Comm'n.; FY 2020 at a Glance, Intellectual Property Import Investigations, U.S. Int'l Trade Comm'n.
14. See Emily Jones, Digital Disruption: Artificial Intelligence And International Trade Policy, 39 OXFORD REV. ECON. POL'Y 70, 71 (2023). Data tagging can invoke serious privacy concerns. Whistleblowers have revealed that human subcontractors manually listen and transcribe audio from Amazon's "Alexa," Google's "Google Home," Apple's "Siri," and Microsoft's "Skype" products to train the AI systems. Paola Tubaro & Antonio A. Casilli, Digital Work in the Planetary Market 175–176 (Mark Graham & Fabian Ferrari eds., 2022).
15. These companies are listed in alphabetical order; the ordering should not be interpreted as a ranking.
16. Emily Jones, Digital Disruption: Artificial Intelligence And International Trade Policy, 39 OXFORD REV. ECON. POL'Y 70, 74 (2023).
17. A total of nine amicus briefs were filed between the merits and petition for en banc rehearing stages in the ClearCorrect appeal before the Federal Circuit. Among amici were the Motion Picture Association of America, Recording Industry Association of America, Association of American Publishers, International Trade Commission Trial Lawyers Association, and International Center for Law and Economics.
18. See ClearCorrect Operating, LLC v. Int'l Trade Comm'n, 810 F.3d 1283, 1286–87 (Fed. Cir. 2015).
19. Chevron, U.S.A., Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837 (1984) was overruled by Loper Bright Enterprises v. Raimondo, 144 S. Ct. 2244 (2024). While the Federal Circuit did not grant Chevron deference to the ITC's interpretation in ClearCorrect, the reader should be aware that going forward, courts will review statutory interpretations by agencies without deference.
20. See 19 U.S.C. § 1337.
21. See ClearCorrect, 810 F.3d at 1287.
22. Id. at 1288.
23. Id.
24. Id. at 1290–94.
25. See id. at 1290–94.
26. Id. at 1300–02.
27. Id. at 1302 (J. O'Malley, concurring).
28. Id. at 1304 (J. Newman, dissenting).
29. Id. at 1305 (J. Newman, dissenting).
30. Id. at 1306, 1310 (J. Newman, dissenting).
31. ClearCorrect Operating, LLC v. Int'l Trade Comm'n, 819 F.3d 1334, 1337–1347 (Fed. Cir. 2016) (denial of petition for en banc rehearing) (J. Newman, dissenting).
32. 17 U.S.C. § 101 (defining a "compilation" as "a work formed by the collection and assembling of preexisting materials or of data that are selected, coordinated, or arranged in such a way that the resulting work as a whole constitutes an original work of authorship").
33. Feist Publ'ns, Inc. v. Rural Tel. Serv. Co., 499 U.S. 340, 358–59 (1991).
34. Id.
35. See ClearCorrect, 810 F.3d at 1286–87.
36. See id.
37. See Suprema, Inc. v. U.S. Int'l Trade Comm'n, 796 F.3d 1338, 1340–41 (Fed. Cir. 2015) (en banc).
38. Id. at 1341–42.
39. Id.
40. Id. at 1343.
41. Introducing R1, A Pocket Companion That Moves AI From Words To Action, Rabbit Inc. (Jan. 9, 2024), https://bit.ly/3Wvy753. The Rabbit R1 product received over $30M in startup funding as of January 2024. See Eric H. Schwartz, Generative AI Hardware Startup Rabbit Reaches $30M in Funding Round, Voicebot.ai (Jan 8, 2024), https://bit.ly/4c8qs2n.
42. The Rabbit R1 is purely used as an example of an AI-first product. The authors are neither suggesting that the product has IP vulnerabilities, nor are the authors predicting that this specific product will be subject to litigation at the ITC or in other venues.
Originally published Westlaw Today
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.