Companies deploying and developing Artificial Intelligence ("AI") face a critical intellectual property dilemma: How should I protect AI trade secrets in a regulatory landscape that demands transparency? Indeed, while the Trade Secrets Act (18 U.S.C. § 1905) criminalizes the unauthorized disclosure of trade secrets or confidential information by federal government personnel, it has an important exception—it applies only to disclosures "not authorized by law." 18 U.S.C. § 1905. On the other hand, federal AI regulations are expected to demand "transparency" and "explainability," requiring businesses to disclose various aspects of their AI technology.1,2 Consequently, new AI regulation mandating transparency may effectively 'authorize by law' the disclosure of AI trade secrets, potentially bringing such disclosures within the exception to § 1905's prohibitions.
With patent and copyright laws still ambiguous in their application to AI, many businesses are turning to trade secrets as their primary shield for AI-related intellectual property.3 Many companies may find it surprising, however, that the strength of trade secret protection in the specific context of regulatory disclosure is not as robust as one may assume.4 For example, under the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA), the EPA may reveal pesticide formulas, even when they contain trade secrets. 7 U.S.C. §§ 136h, 136j. The landscape of trade secret disclosure in federal regulation is complex, with various laws either permitting or prohibiting disclosure under different circumstances.5,6
A key principle from case law is that confidential information shared with government agencies is not automatically protected; rather, the protection depends on specific assurances of confidentiality and the context of disclosure. As an example, the EPA's authority under FIFRA to disclose pesticide formulas, as mentioned above, was at the center of Ruckelshaus v. Monsanto Co., a Supreme Court case that shaped the understanding of trade secret protection in regulatory contexts.
In Ruckelshaus, the Supreme Court emphasized the importance of expectations of confidentiality when sharing trade secrets with government agencies. The Court held that statutory assurances that confidential information will be kept secret create a reasonable expectation of protection of that information.7 However, it also noted that when laws clearly outline potential disclosure, companies cannot reasonably expect confidentiality. As the Court stated, "As long as [the company] is aware of the conditions under which the data are submitted, and the conditions are rationally related to a legitimate Government interest, a voluntary submission of data by an applicant in exchange for the economic advantages of a registration can hardly be called a taking."8 This ruling underscores that, by default, companies should not assume their confidential information will be protected when shared with government agencies unless explicit assurances are given or statutory protections are in place.
Following Ruckelshaus, which dealt with agency disclosure of trade secrets, courts have grappled with similar issues in the context of Freedom of Information Act (FOIA) requests, providing additional insights into the balance between confidentiality and transparency.
In Food Marketing Institute v. Argus Leader Media (2019), the Supreme Court established that, in the context of an FOIA request, information is confidential when it is "both customarily and actually treated as private by its owner and provided to the government under an assurance of privacy."9 This ruling eliminated the previous requirement to demonstrate substantial competitive harm, thereby strengthening protections for confidential information.
Building on the principles established in Argus Leader, in Flyers Rights Education Fund v. Federal Aviation Administration (2023), the D.C. Circuit Court clarified that broad promises of "transparency" or "open and honest communication" by agencies or corporations do not constitute an explicit representation that confidential information will be disclosed and therefore do not destroy a company's expectation of confidentiality.10
These cases illuminate a critical yet sometimes overlooked aspect of trade secret protection in the regulatory landscape: the absence of automatic safeguards for confidential information shared with federal agencies. This revelation may surprise many in the AI industry who assume their trade secrets are inherently protected when complying with federal regulations.
As we look toward the future of AI regulation, it's crucial to understand how these legal precedents might apply. The US Senate AI Working Group's policy roadmap11 indicates that AI regulation will likely occur through existing regulatory frameworks and agency enforcement, with targeted legislative updates to address specific gaps. This approach is exemplified by the working group's statement: "We encourage the relevant committees to consider identifying any gaps in the application of existing law to AI systems that fall under their committees' jurisdiction and, as needed, develop legislative language to address such gaps."12 This approach means AI companies must navigate a complex landscape where the protection of their trade secrets will depend on the specific regulatory context.
Based on the existing legal framework and anticipated AI regulations, companies face three potential scenarios regarding the disclosure of their confidential information:
- Permissive Disclosure: Some regulations may allow agencies to disclose trade secrets under certain circumstances.
- Prohibited Disclosure: Other regulations may explicitly bar agencies from revealing certain types of information.
- Mandated Disclosure: Emerging AI-specific regulations focused on transparency and explainability may require companies to disclose information they consider trade secrets.
The challenge for AI companies lies in anticipating which scenario they might face and preparing accordingly. As the AI regulatory environment evolves, with its emphasis on transparency and explainability, the balance between innovation protection and regulatory compliance will become increasingly complex.
Companies venturing into regulated AI territories should:
- Carefully examine applicable laws and regulations.
- Seek explicit confidentiality assurances from federal agencies where possible.
- Prepare strategies for each potential disclosure scenario.
- Stay informed about emerging AI-specific regulations and their potential impact on trade secret protection.
By understanding the legal precedents set by cases like Ruckelshaus, Argus Leader, and Flyers Rights, and applying these lessons to the emerging AI regulatory landscape, companies can better navigate the challenges of protecting their innovations while meeting regulatory requirements.
Footnotes
1 See e.g., "Driving U.S. Innovation in Artificial Intelligence: A Roadmap for Artificial Intelligence Policy in the United States," U.S. Senate (May 2024), https://www.schumer.senate.gov/imo/media/doc/Roadmap_Electronic1.32pm.pdf;Executive Office of the President, "Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence," 88 Federal Register 75191, November 1, 2023, at https://www.federalregister.gov/documents/2023/11/01/2023-24283/ safe-secure-and-trustworthy-development-and-use-of-artificial-intelligence; White House, Blueprint for an AI Bill of Rights: Making Automated Systems Work for the American People, October 2022, at https://www.whitehouse.gov/wp-content/uploads/2022/10/Blueprint-for-an-AI-Bill-of-Rights.pdf; Nicol Turner Lee, "Making AI more explainable to protect the public from individual and community harms," Brookings Institution (Nov. 29, 2023) https://www.brookings.edu/articles/making-ai-more-explainable-to-protect-the-public-from-individual-and-community-harms/; Bill Whyman, "AI Regulation is Coming- What is the Likely Outcome?" Center for Strategic & International Studies (Oct. 10, 2023) https://www.csis.org/blogs/strategic-technologies-blog/ai-regulation-coming-what-likely-outcome.
2 Both International and U.S. state AI regulations depict the regulatory trend demanding AI transparency and explainability. For example, the EU AI Act requires extensive documentation of AI systems and models describing the operations of the same. See European Union's Artificial Intelligence Act, Regulation (EU) 2024/1689 ("EU AI Act"). The AI laws passed in Colorado and Utah have similar regulatory objectives. See Colorado Senate Bill 24-205, Concerning Consumer Protections in Interactions with Artificial Intelligence Systems, signed May 17, 2024, effective February 1, 2026; Utah State S.B. 149, the Artificial Intelligence Policy Act (the AI Act), signed Mar. 13, 2024, effective May 1, 2024.
3 See e.g.,Lauren Castle, "Trade Secrets Summoned to Protect AI Amid Noncompete Uncertainty," Bloomberg Law (July 16, 2024) https://news.bloomberglaw.com/ip-law/trade-secrets-summoned-to-protect-ai-amid-noncompete-uncertainty; Jeremy Elman, "AI and Trade Secrets: A Winning Combination," IP Watchdog (Nov. 28, 2023).
4 See Camilla Hrdy, "Morten Follow-Up: What Do Federal Agencies' Enabling Statutes Say About Their Power to Disclose Trade Secrets?" Written Description (Sept. 6, 2022) https://writtendescription.blogspot.com/2022/09/morten-follow-up-what-do-federal.html; see also Morten, Christopher, Publicizing Corporate Secrets (December 1, 2023). University of Pennsylvania Law Review, Vol. 171 (2023), Available at SSRN: https://ssrn.com/abstract=4041556; https://writtendescription.blogspot.com/2022/08/christopher-morten-do-federal-agencies.html.
5 Permissive statute examples: National Transportation Safety Board (NTSB): Title 49 permits NTSB to disclose accident data, even if it includes trade secrets (49 U.S.C. § 1114(b)(3)) ; Environmental Protection Agency (EPA): FIFRA permits EPA to disclose pesticide formulas (7 U.S.C. §§ 136h, 136j); National Institutes of Health (NIH): Food and Drug Administration Amendments Act implicitly permits NIH to disclose clinical trial data (42 U.S.C § 282(j)); Food & Drug Administration (FDA): Food, Drug, & Cosmetic Act permits FDA to disclose clinical trial data (21 U.S.C. §§ 371(a), 393(b)).
6 Prohibitive statute examples: FDA: barred from revealing drug production methods (21 U.S.C. § 331(j)); Federal Trade Commission (FTC): restricted from disclosing trade secrets (15 U.S.C. § 46(f)); U.S. Patent and Trademark Office (USPTO): barred from disclosing patent applications under certain situations (35 U.S.C. § 122).
7 Ruckelshaus v. Monsanto Co., 467 U.S. 986 (1984).
8 Id. at 1007-1008.
9 Food Marketing Institute v. Argus Leader Media, 588 U.S. 427, 440 (2019) (emphasis added).
10 Flyers Rts. Educ. Fund, Inc. v. Fed. Aviation Admin., 71 F.4th 1051, 1055-56 (D.C. Cir. 2023).
11 "Driving U.S. Innovation in Artificial Intelligence: A Roadmap for Artificial Intelligence Policy in the United States," U.S. Senate (May 2024), https://www.schumer.senate.gov/imo/media/doc/Roadmap_Electronic1.32pm.pdf.
12 Id.
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.