AI shockwave to come in trade secret disputes

Just as practitioners may feel adjusted to the surge of trade secrets litigation over the last decade, artificial intelligence (AI) is reshaping the landscape in ways that will likely prompt companies and courts to rethink long-standing assumptions.

The Defend Trade Secrets Act, enacted in 2016, opened the door to federal trade secret claims, accelerating an already growing wave of high-stakes disputes – especially after the Federal Circuit’s 2011 TianRui decision boosted the use of the International Trade Commission for such cases.

The consequential growth in both the number and size of trade secrets cases over the past decade was driven by a number of factors, including (1) the speed of technology, which may render patenting too slow for meaningful protection and (2) the ever-increasing interconnections among businesses.

In this alert, we discuss how AI is expected to change several aspects of trade secrets litigation.

Proof of trade secrets

For one thing, AI complicates the process of proving that a trade secret exists in the first place. Fundamental to a trade secrets claim, the plaintiff must prove that the asserted trade secret is actually secret, and that it derives value because of that secrecy.

However, as use of AI tools and other models became widely available over the last several years, employees and researchers inevitably used them to develop or refine proprietary information (whether policy allowed them to do so or not). Notably, both prompts and outputs are stored by the AI platform and, in some cases, can be used to train or fine tune the AI model itself. As a practical matter, this function risks leaking the secrecy of the information, therefore undermining any trade secrets claim. Inevitably, as engineers and business teams turn to AI for their daily work, inadvertent leaks pose a grave risk.

Challenging trade secret violations

This also means that AI will likely make trade secrets easier to challenge. While trade secrets cannot be generally known, they also cannot simply reflect information that can be easily reverse-engineered.

However, AI models are increasingly capable of inferring hidden information from public sources at incredible speeds. This can be demonstrated by feeding a redacted document into a preferred AI model and asking it to guess what lies under the redaction.

Competitors are likely to feed patents, publications, product data, and even redacted court documents into AI tools and ask them to “fill in the gaps.” Trade secrets case defendants are expected to use this type of AI-enabled reverse-engineering to argue that a claimed trade secret was easily discoverable all along.

Discovery

AI is expected to trigger new discovery battles in litigation. Because AI use may reveal or compromise purported secrets, parties could seek extensive discovery on each other’s AI activity – policies, tools, prompts, outputs, logs, and historical usage – including employees’ and contractors’ usage at home, on their personal devices.

Courts long ago developed frameworks for e-discovery, but disputes over AI-related data – some of it years old – are set to introduce new layers of complexity. Discovery is likely to spread to third parties, including the sponsors of the AI models themselves. It is difficult to predict where any particular court will land on the scope of such discovery.

Patenting

Those responsible for managing corporate intellectual property (IP) will also see a significant AI shift in the patent–trade secret calculus. Companies carefully weigh what to patent and what to keep confidential, considering timing, value, and the risk of disclosure; AI disrupts all of these factors. It is set to enable competitors to deduce inventions almost as quickly as applications can be drafted, and it will likely shorten the commercial lifespan of innovation by accelerating competitive insight in many cases. Organizations that depend on their IP are encouraged to begin rethinking how they evaluate, protect, and exploit their IP in an AI-driven environment. This will include careful weighing of the value of IP in an environment where AI can not only help design-around patents, but also guess at unpatented trade secrets.

Employee disputes

AI will also likely reshape employee mobility and “memory” disputes. Employee departures already trigger litigation over what information an employee allegedly misappropriated as part of their departure. AI tools – especially those integrated into workplace systems – blur the line between an employee’s personal know-how and company-owned information. The more employees rely on AI copilots to draft code, design systems, or generate business strategies, the more questions could emerge about whether the AI outputs were effectively stored, logged, or traceable long after the employee leaves. Litigation over whether an employee’s use of AI expanded, preserved, or unfairly transferred knowledge that would otherwise exist only in human memory – and who owns the human memory that remembers how to prompt the AI – is expected to increase.

AI will likely also create new categories of “derived” or “model-dependent” trade secrets. Historically, trade secrets have involved human-created information: formulas, processes, customer data, strategies. However, as companies build proprietary AI models – and as those models generate insights, patterns, and optimizations – courts will have to confront whether model-generated knowledge (and the models themselves) qualify as a protectable trade secrets. Disputes will likely arise over whether the weights, training data selections, embeddings, or emergent insights of proprietary models constitute trade secrets, and whether defendants can be liable for misappropriation when they never accessed the underlying inputs directly. Companies and practitioners can expect this new frontier to test the limits of traditional trade secret doctrines.

Ordinary skill

Finally, AI is going to change the “ordinary skill in the art.” Until now, a skilled practitioner for patenting purposes was one with training in the relevant field and likely a computer to assist their work. Now, however, a person of ordinary skill has training and can also be presumed to have the then-latest AI model to assist them in understanding the prior art and making ordinary improvements to devices and processes. Will this raise the bar of what is obvious? Does it render a patent specification’s written description and enablement to include whatever an AI would understand from the disclosure? These are open questions – but insofar as they affect the ability to effectively create enforceable patents, they also change the calculus of what to keep unpatented and secret.

Conclusion

AI is opening new opportunities, but it’s also creating new risks for how companies develop, store, and protect their most valuable information. Organizations are encouraged to review their IP strategies, clarify internal AI‑use policies, and ensure that safeguards keep pace with the technology.

To avoid the risk of losing cases to more agile competitors, parties in trade secrets cases and their attorneys would need to apply thoughtful planning. With preparation, companies and their attorneys can stay ahead of emerging legal challenges and continue to protect the innovations that drive their businesses.

For more information, please contact the authors.