Protecting Trade Secrets in the Era of Generative Artificial Intelligence – Part II: Novelty

Protecting Trade Secrets in the Era of Generative Artificial Intelligence – Part II: Novelty

Last month we discussed how the rapid adoption of ChatGPT and its work-related usage may create new risks for businesses regarding the confidentiality requirements to establish trade secret protection. This article highlights another way that generative AI may change the trade secret landscape:  it may redefine what constitutes novelty and how to prove it.

Under Washington law, a party seeking trade secret protection must demonstrate that the purported secrets are novel.[1] Novel ideas that constitute trade secrets could include, for example, business strategy concepts, production methods or techniques, or recipes and formulas. “To be novel, ‘the information must not be readily ascertainable from another source.’”[2] Proving novelty requires showing that the alleged trade secret cannot be easily reproduced from publicly available information.[3] For example, the Ninth Circuit has held that recipes developed by a buffet style restaurant for “American staples” such as “macaroni and cheese” were not entitled to trade secret protection because they were readily ascertainable and could be easily reproduced.[4]

Generative AI models, such as OpenAI’s GPT-4, are capable of producing a vast array of outputs in human-like text. As AI tools improve, they could potentially generate ideas for products and strategies similar to those humans might otherwise conceive. Indeed, this is a key value proposition of generative AI: that it can learn patterns from a large quantum of data, and then generate outputs that may mimic the substance, style, and complexity of human creativity. This could, as both AI and the law evolve, present a challenge for businesses: how can they prove their product feature, strategy plan, formula or methodology is novel when it can be replicated by AI through common-sense prompts?

Perhaps novelty will be found in the methods used to manipulate AI tools as part of a larger process, or in the particulars of the inputs utilized. It seems certain, however, that the time is now for businesses to assess what comprises their competitive advantages, and how best to leverage and protect those advantages in the face of rapidly evolving AI technology.

Brandi B. Balanda


This is the second article in our series “Mitigating Legal Risk in Today’s Business Climate – a Litigator’s Lens.”


[1] RCW 19.108.010(4)(a); Inteum Co., LLC v. Nat’l Univ. of Sing., 371 F.Supp.3d 864, 877 (W.D.Wash. 2019); Barrett Bus. Servs., Inc. v. Colmenero, 2022 U.S. Dist. LEXIS 227966, *12 (E.D. Wash. Dec. 19, 2022).

[2] Inteum Co., LLC, 372 F.Supp.3d at 877 (quoting Robbins, Geller, Rudman & Dowd, LLP v. State, 179 Wn. App. 711, 328 P.3d 905, 911 (Wn. App. 2014)).

[3] Id.

[4] Buffets, Inc. v. Klinke, 73 F.3d 965 (9th Cir. 1995).