I. Introduction
Artificial intelligence is reshaping not just industries but the legal tools companies use to protect their innovations. While patents and copyrights have long anchored intellectual property (IP) strategies for tech companies, their limitations are becoming increasingly apparent in the face of rapid AI-driven change. As companies confront mounting uncertainty over what can be patented or copyrighted—and how reliably those rights can be enforced—the focus is shifting toward trade secrets as the most adaptable and pragmatic means of safeguarding proprietary advantage. This article explores why, despite the continued importance of patents and copyrights, trade secrets now stand at the forefront of IP strategy in the AI era. We examine the legal, technical, and operational challenges unique to protecting innovations shaped or enabled by AI, and outline how companies can build an effective, resilient IP program centered on secrecy.
II. How Patent and Copyright Protections Are Further Eroding in the AI Age
A. Patents: Section 101 and the Uncertain Status of AI-Assisted Inventions
For decades, innovation-driven businesses have relied heavily on two pillars of intellectual property law: patents and copyrights. But over time, a combination of judicial retrenchment, shifting regulatory interpretations, and technological disruption has steadily weakened these protections—especially in the software and now AI sectors.
The decline of software and AI patents as reliable commercial assets can be traced to the aftermath of the Supreme Court’s 2014 decision in Alice v. CLS Bank, where the Court invalidated a financial software patent under 35 U.S.C. § 101, holding that implementing an abstract idea on a computer does not make it patent-eligible.1 In the years since, software patent applications—particularly those involving AI—have faced rising rejection rates, often on the grounds that they merely describe abstract algorithms not patentable because they can be done within the human mind and are part of the “‘basic tools of scientific and technological work’ that are open to all.”2
Generative AI presents a paradox: it accelerates innovation while simultaneously undermining the ability of individual inventors and creators to protect it as intellectual property.
Compounding the problem, even at this early stage of the AI era, it is safe to assume that AI will play some role in virtually all inventive processes going forward. And even when such AI-assisted inventions clear the § 101 patent-eligible subject matter hurdle, they face a second, growing challenge: meeting the new “significant human contribution” threshold for inventorship. In February 2024, the USPTO issued guidance extending the Pannu joint inventorship framework to cover AI-assisted inventions, requiring applicants to show that a named human inventor meaningfully contributed to the conception of at least one claim.3 This adds not only a substantive burden but also a disclosure obligation: inventors must now provide detailed records tracing human contributions alongside expansive machine-generated inputs.4
B. Copyrights: AI-Assisted Coding and the Limits of Copyrightability
Copyright law, once a cornerstone tool for software companies, has also diminished in relevance. The first impetus for this shift was the rise of the software-as-a-service (SaaS) model. Under SaaS, companies rarely distribute executable or object code to customers; instead, they provide access to hosted services. As such, the practical role of copyright protections over software has diminished because under the SaaS model, customers no longer receive copies of the executable code. This reduced the risk of customer-side infringement—because customers no longer receive or control the code—and made traditional anti-piracy enforcement mechanisms less central to software protection strategies.5
The decline and fall of copyright for software code is further accelerated in the AI era, where most software development is, or soon will be, AI-assisted. As developers increasingly use tools like GitHub Copilot or ChatGPT to write and refactor code, the copyright eligibility and enforceability of resulting outputs become progressively more uncertain.
One reason is that generative AI tools may incorporate snippets of open-source code—sometimes subject to copyleft or other restrictive licenses—into developer outputs. This can inadvertently transform proprietary software into code subject to open-source obligations, undermining exclusive rights and creating unforeseen licensing and distribution requirements.
More fundamentally, the U.S. Copyright Office (USCO) has made clear that copyright protects only the human-authored aspects of a work.6 Determining which parts of a work are “human-authored”—particularly when the line between human prompt and machine output blurs—raises thorny legal and evidentiary challenges. The standards for establishing which portions reflect human creative decisions and how much human input is sufficient to qualify for authorship under copyright law have yet to be established. Moreover, under the USCO’s disclosure framework, applicants must report any inclusion of AI-generated material and explain the human author’s contributions, adding a compliance burden during the application process and creating additional grounds for invalidation in any future enforcement action. These challenges are not just technical hurdles; they fundamentally reshape how software companies must approach IP strategy, risk management, and competitive positioning in the AI era.
In sum, while patents and copyrights remain important tools, the challenges of securing and enforcing them in the AI age significantly reduce their practical weight as standalone protections. Increasingly, companies are supplementing these traditional IP categories by turning to the last major pillar of innovation protection: trade secrets.
In Part 2, we’ll break down the legal, technical, and policy safeguards essential for building resilient trade secret programs in the AI age—and explain why secrecy has become not just a tactical choice but a strategic imperative.
著者について
Jim W. Ko is a patent attorney and focuses his practice on providing counsel for all the ways that intellectual property and artificial intelligence issues can and will impact businesses. He lives in Chandler, Arizona.
