Trade Secret Law in the AI Era: How Bot-Based Scraping, DTSA Pleading Standards, and Unjust Enrichment Are Reshaping the Doctrine

Thursday, December 4, 2025 

Executive Summary

Instructor(s)
Camilla Hrdy, Rutgers Law School
Rajiv Patel, Fenwick
Thomas Sprankling, WilmerHale
Yar Chaikovsky, White & Case

Keywords

Defend Trade Secrets Act (DTSA) — pleading standard — sufficient particularity vs. reasonable particularity • CompuLife Software, Inc. v. Newman — Eleventh Circuit — bot scraping as improper means • Quintara Biosciences v. Ruifeng Biztech — Ninth Circuit — DTSA identification timing • Insulet Corp. v. EOFlow Co. — DTSA statute of limitations — discovery rule vs. inquiry notice • Snyder v. Beam Technologies — Tenth Circuit — trade secret standing — possession vs. ownership • Hytera Communications Corp. v. Motorola Solutions — Seventh Circuit — DTSA extraterritorial act in furtherance • Trade secret unjust enrichment damages — apportionment requirement — DTSA vs. patent law • AI-generated trade secrets — no human-creator requirement — DTSA ownership and standing • Kewanee Oil Co. v. Bicron Corp. — federal patent preemption of state trade secret law • DTSA extraterritoriality — 18 U.S.C. 1837 — act in furtherance standard • “can a bot scraping publicly available data constitute trade secret misappropriation” • “what is the difference between DTSA and California CUTSA pleading standards for trade secrets”

Legal Analysis

AI-Generated Trade Secrets, Bot-Based Misappropriation, and Whether CompuLife’s “Readily Ascertainable” Analysis Survives the Age of LLMs

The trade secret law of 2025 is being reshaped by three intersecting developments: the enactment and maturation of the Defend Trade Secrets Act, Pub. L. No. 114-153, 130 Stat. 376 (2016), which created the first federal civil cause of action for trade secret misappropriation; the progressive weakening of patent eligibility under 35 U.S.C. § 101 and enforceability through the IPR regime, which has pushed companies toward trade secret protection as an alternative; and the emergence of large language models and AI-based analytical tools, which have destabilized the doctrinal assumption that information requiring significant human effort to compile is not “readily ascertainable through proper means.” Hrdy reported that approximately 1,400 federal trade secret cases are filed annually since the DTSA’s passage — a level that spiked immediately after 2016 and has since plateaued — with an 80-plus-percent plaintiff win rate at trial, though the vast majority of cases settle or resolve on early injunction. The damages landscape shows a median award under one million dollars but an average exceeding eight million dollars, driven upward by a subset of very large verdicts that, as Chaikovsky observed, “rival or dwarf patent cases” and are increasingly founded on unjust-enrichment theories that face none of the Federal Circuit’s judicially imposed apportionment discipline.

The definitional question that AI most acutely disrupts is “readily ascertainable through proper means” — a threshold element under the DTSA and most state uniform trade secret laws that functions as a proxy for meaningful secrecy without requiring the plaintiff to prove novelty or non-obviousness in the patent-law sense. CompuLife Software, Inc. v. Newman, 959 F.3d 1288 (11th Cir. 2020), and its follow-on proceedings, constitute the leading federal appellate treatment of automated data collection as both a readily-ascertainable and an improper-means question. In CompuLife, the defendants hired a programmer to build a bot that automatically extracted life insurance quotes from the plaintiff’s website and compiled them into a competing dataset; the Eleventh Circuit held — after extensive litigation — that this constituted misappropriation because, while a human could theoretically have recorded each individual quote manually, using a bot to automate that extraction at scale constituted improper means and the resulting compiled dataset was not readily ascertainable through proper means. Hrdy described CompuLife as “really one of the most important trade secret cases since DuPont v. Christopher,” referring to E.I. duPont de Nemours & Co. v. Christopher, 431 F.2d 1012 (5th Cir. 1970), in which aerial surveillance of a chemical plant from public airspace was held to constitute improper means on the rationale that the law provides a remedy whenever “the holder of a trade secret would otherwise be required to protect himself by extraordinary means.” The structural parallel — a human could have obtained the same information by lawful means, but the automated method crosses the line — is now the central unsettled question in AI-era trade secret law.

The panel identified two doctrinal complications that make CompuLife’s holding fragile as a forward-looking rule. First, as Patel observed, the plaintiff in CompuLife did not initially have terms of service prohibiting automated access — adding them only after litigation commenced — and courts frequently look to terms-of-service prohibitions in assessing both improper means and whether reasonable measures were taken to maintain secrecy. Second, Spranklen raised what he called a Kyllo-style analytical framework — drawing on Kyllo v. United States, 533 U.S. 27 (2001), in which the Supreme Court held that using thermal imaging technology not in general public use constituted a Fourth Amendment search — as a potential alternative to the human-could-have-done-it test: if reasonable expectations of information security shift as AI becomes ubiquitous, the threshold for what counts as readily ascertainable may shift with it. Chaikovsky pressed the practical implication: if an LLM can reconstitute from public sources what a company believes is a proprietary compilation, the protection that CompuLife’s bot-equals-improper-means rule provides may erode as the technology democratises. Patel’s practice recommendation in light of this uncertainty was risk-layering: companies should document their reasonable-measures efforts at every stage, because “if you don’t have any of that, then you could run into a case where you just don’t know where it’s gonna open up into, given that it’s still forming as law.” The Supreme Court’s denial of certiorari in CompuLife left the Eleventh Circuit’s analysis as the most authoritative available federal treatment — binding nowhere else, persuasive everywhere.

DTSA Pleading Particularity After Quintara, the California Pre-Discovery Identification Rule, and Statute of Limitations Splits in Insulet and Snyder

The question of when a plaintiff must identify its trade secrets with sufficient specificity — at the pleading stage, before discovery, or only at summary judgment — has become the front-line practical battleground in trade secret litigation, with direct consequences for both case duration and the cost of defense. California’s Uniform Trade Secrets Act, Cal. Civ. Code § 2019.210, requires a plaintiff to identify its claimed trade secrets with reasonable particularity before the discovery it seeks to compel may begin, a rule whose stated policy rationale is preventing plaintiffs from using discovery as a fishing expedition to define their secrets by examining the defendant’s technology after the fact. The Ninth Circuit’s recent decision in Quintara Biosciences, Inc. v. Ruifeng Biztech Inc., [9th Cir. 2025; full citation not in transcript], authored by Judge Johnstone, held that the DTSA imposes a different and apparently more plaintiff-permissive standard — “sufficient particularity” rather than “reasonable particularity” — and that the identification requirement under the DTSA need not be satisfied before discovery commences but may be deferred to the summary judgment stage. Hrdy assessed the outcome as doctrinally defensible on the statutory text: “the DTSA doesn’t say anything about” pre-discovery identification, whereas California’s statute does, and the court’s reading correctly declines to import a California-specific procedural rule into federal law. Spranklen reserved his assessment of whether the outcome is correctly decided while noting its immediate practical effect: early motions to dismiss for failure to identify trade secrets are now substantially harder to prosecute in the Ninth Circuit, and the suggestion that a “very quick summary judgment decision” may substitute is, in Chaikovsky’s assessment, cold comfort for defendants facing the cost of full discovery on an ill-defined IP right.

The statute of limitations produced a parallel circuit split that the panel examined through Insulet Corp. v. EOFlow Co., Ltd., [D. Mass. 2024–2025; full citation not in transcript]. The DTSA provides a three-year limitations period running from the date the plaintiff “discovered the misappropriation or by the exercise of reasonable diligence should have discovered the misappropriation.” In Insulet, a former employee left to join a competing medical device company, which then introduced a patch insulin pump product that bore visual similarities to the plaintiff’s product, demonstrated at an industry conference. The defendant argued that the prototype demonstration, combined with the employee departure, was enough to start the limitations clock; the court declined, holding that “mere suspicion” is insufficient and that a plaintiff is entitled to give a competitor a degree of good faith — effectively requiring something closer to actual knowledge of misappropriation rather than mere suspicion. Spranklen observed that the Insulet court acknowledged it was applying a standard that “no other court interpreting the DTSA” had previously adopted, and drew its formulation from Merck & Co. v. Reynolds, 559 U.S. 633 (2010), a securities fraud case, rather than from established trade secret precedent. His practice advice was categorical: “if you’re a plaintiff, file your case early” and never rely on a favorable statute-of-limitations ruling because the law remains in flux across circuits. The companion ownership question was addressed in Snyder v. Beam Technologies, Inc., [10th Cir. 2025; full citation not in transcript], where an employee who had carried a client list from his former employer to a new employer was held to potentially have standing to sue his new employer for misappropriating that list under state trade secret law — even though he had himself taken the information from his prior employer — because the applicable state law required only possession, not ownership; the DTSA’s requirement of ownership, by contrast, would have denied standing. Hrdy described the outcome as “patently ridiculous” as a matter of policy, attributing it to the divergence between the DTSA’s ownership-based standing requirement and most states’ possession-based standing rules.

Trade Secret Damages Without Apportionment, DTSA Extraterritoriality Under Hytera, and the IP Layering Strategy as Structural Response

Trade secret damages are governed by a framework that is less doctrinely constrained than the Federal Circuit’s patent damages jurisprudence in two structurally significant respects. First, apportionment — which the Federal Circuit has rigorously required in patent cases, demanding that plaintiffs isolate the incremental value of the patented feature from the value of the entire accused product — has no clearly established equivalent in trade secret law, with courts split on whether there is an apportionment requirement at all, who bears the burden of production, and how it should be measured. Spranklen described the practical consequence for defendants: “some people have gone so far as to argue it’s enough to say your product has my trade secret in it” and “now you need to tell the court how much of it is not attributable to the trade secret,” a burden-shifting formulation that, if adopted, would functionally replicate in trade secret law the pro-patentee position that the Federal Circuit explicitly rejected in its damages reform era. Second, unjust enrichment as an alternative to reasonable royalty or lost profits provides plaintiffs a damages theory whose ceiling can encompass the entire profit attributable to the misappropriated information — a figure that Chaikovsky noted “tends to come from the unjust enrichment component” in the largest verdicts and is “a lot harder to fight back” at the appellate level because the factual record supporting disgorgement is typically less susceptible to legal error than a royalty-rate calculation. Adding a RICO claim to a DTSA case where multiple conspirators are involved, Chaikovsky observed, “you are in the mix automatically” for treble damages, producing an exposure that can escalate to multi-billion dollar territory without the moderating effect of Federal Circuit scrutiny.

The DTSA’s extraterritorial reach has been significantly expanded by a line of cases interpreting 18 U.S.C. § 1837’s “act in furtherance” requirement. The statute applies where the defendant is a United States citizen or where “an act in furtherance of the offense was committed in the United States.” In Motorola Solutions, Inc. v. Hytera Communications Corp., 495 F. Supp. 3d 687 (N.D. Ill. 2020), aff’d in relevant part, [7th Cir.], the Seventh Circuit held that an act in furtherance does not require a consummated act of misappropriation in the United States — a single step taken within the domestic territory in support of a broader scheme of misappropriation located primarily abroad can satisfy the jurisdictional predicate. Chaikovsky reported that the trend in act-in-furtherance rulings “seem to be going more in the Hytera direction” and that the practical effect is to make “extraterritoriality” available in virtually every DTSA case with any US nexus, because even employee poaching from a California company, or post-misappropriation US product sales, has been found sufficient. Spranklen noted the structural anomaly: the DTSA’s express extraterritoriality provision in 18 U.S.C. § 1837, hooked to the Economic Espionage Act of 1996, effectively overrides the presumption against extraterritoriality — “a statute that expressly says it, which I think is fairly unusual” — and that the Supreme Court has not yet provided guidance on the outer limits of “act in furtherance,” having declined certiorari in CompuLife and never having taken a DTSA case. The only Supreme Court precedent on trade secret preemption remains Kewanee Oil Co. v. Bicron Corp., 416 U.S. 470 (1974), which held that federal patent law does not preempt state trade secret law — a foundational but now architecturally insufficient anchor for the system the DTSA has built atop it.

Against this background of doctrinal uncertainty, the panel converged on an IP-layering strategy as the most robust structural response available to practitioners. Patel described a case in which a company protecting software as a trade secret simultaneously prosecuted a patent application covering the functionality but not the underlying implementation details, accelerated prosecution after an employee misappropriated the code, and settled the resulting litigation more quickly because “the other side just didn’t wanna deal with” the combined trade secret and patent exposure. The key premise of layering is that each protection regime — patent, copyright, and trade secret — covers a different dimension of the same underlying innovation: patent protects functional methods and structures publicly disclosed, copyright protects the literal expression of code once registered using the Copyright Office’s partial-deposit provisions that preserve trade secret in the non-deposited portion, and trade secret protects the proprietary implementation details that neither regime discloses. Spranklen and Patel both flagged the critical caution that Spranklen described as requiring unambiguous emphasis: “you should not” include your trade secret in a patent application, because patent applications are published eighteen months after filing under 35 U.S.C. § 122(b), automatically destroying the secrecy on which trade secret protection depends. Hrdy’s closing doctrinal observation was that the AI era will place increasing pressure on trade secret law specifically because AI-generated works cannot be protected by patent (requiring a human inventor) or copyright (requiring sufficient human authorship), yet can be protected as trade secrets because trade secret law has no human-creator requirement — a structural asymmetry that she predicted will drive a significant expansion of trade secret assertion in the coming years.

Generated by AI based on the Interview/Transcript below.

Key Takeaways

• AI Output Can Be a Trade Secret Without a Human Creator: Unlike patent and copyright law, trade secret law has no human-creator requirement; Spranklen confirmed that “AI could create a trade secret” that is a “legally enforceable right,” while Hrdy predicted this will put increasing pressure on trade secret doctrine as AI-generated works proliferate beyond the reach of patent and copyright protection.
• Quintara’s “Sufficient Particularity” Standard Favors Plaintiffs: The Ninth Circuit’s Quintara decision holds the DTSA imposes a “sufficient particularity” standard distinct from California’s pre-discovery “reasonable particularity” rule, allowing plaintiffs to defer full trade secret identification to the summary judgment stage; Chaikovsky warned this “necessarily” drives litigation costs up significantly because defendants cannot use early identification to pare the case down.
• Bot Scraping Likely Constitutes Improper Means: CompuLife Software, Inc. v. Newman (11th Cir. 2020) held that using a bot to compile data that a human could have assembled manually — but that required automated extraction at scale — constituted both improper means and a viable trade secret compilation claim; Hrdy described CompuLife as one of the most important trade secret cases since E.I. duPont de Nemours & Co. v. Christopher, though the Supreme Court denied certiorari, leaving the rule binding only in the Eleventh Circuit.
• Trade Secret Damages Are Not Apportionment-Constrained: Unlike patent damages, trade secret awards face no clearly established apportionment requirement; unjust enrichment claims can capture total defendant profits attributable to the trade secret, and DTSA-plus-RICO cases can produce treble damages, creating what Chaikovsky described as a damages environment where “numbers are big” and appellate courts have not yet developed the moderating discipline the Federal Circuit applies to patent verdicts.
• DTSA Extraterritoriality Is Expansive Under Hytera: Motorola Solutions, Inc. v. Hytera Communications Corp. established that a single US-based step in a broader foreign misappropriation scheme satisfies the DTSA’s “act in furtherance” requirement; Chaikovsky cautioned that this makes multi-billion-dollar damages claims effectively available in nearly any DTSA case with minimal US contact, and the Supreme Court has not yet addressed the outer limits.
• Possess, Don’t Disclose: The Layering Imperative: Patel’s core strategic recommendation is IP layering — simultaneously protecting software through trade secret (implementation), patent (function), and copyright (expression via partial-deposit registration) — with the absolute constraint that Spranklen stated as non-negotiable: “you should not” include trade secrets in a patent application, because publication under 35 U.S.C. § 122(b) destroys secrecy.
• Standing to Sue Requires Ownership Under DTSA, Not State Law: Snyder v. Beam Technologies (10th Cir. 2025) illustrates that under most state uniform trade secret laws, a mere possessor — including one who derived possession through theft from a prior employer — may have standing to sue, while the DTSA requires ownership; Hrdy described the state-law result as “patently ridiculous” but consistent with statutory text.
• File Early: Statute of Limitations Remains Unsettled: Insulet Corp. v. EOFlow held that mere suspicion of misappropriation — including a visually similar competitor prototype and former-employee departure — is insufficient to start the DTSA’s three-year limitations clock, but Spranklen’s universal advice was: “if you\’re a plaintiff, file your case early,” because the discovery-rule standard under the DTSA remains unsettled across circuits and California applies the more demanding diligence standard.
• AI Scraping and Terms of Service: Protect Proactively: The absence of a terms-of-service prohibition on automated access was a material weakness in CompuLife’s position; Hrdy confirmed courts look to whether scraping was “prohibited by the terms of use” in assessing both improper means and reasonable measures, making robust ToS drafting a front-line trade secret defense mechanism.

B-CLE Recording (CLE: $100) | Resource(s) | Speaker Bio(s) & Contact Info

Download the Panel Transcript and Materials here!

Interview/Transcript

This interview/transcript was based on a conversation on December 4, 2025 at the 26th Annual Berkeley-Stanford Advanced Patent Law Institute. This transcript addresses Hot Topics in Trade Secrets, featuring speakers Camilla Hrdy, Rutgers Law School, Rajiv Patel,Fenwick, and Thomas Sprankling, WilmerHale, and moderated by Yar Chaikovsky, White & Case. 

[WAYNE]

So I wanted to welcome everybody back for this session. Uh, I think this is an interesting session in terms of the progression of the Advanced Patent Law Institute.

 

We go back and look at it over the last 20 years, trade secret has only emerged in the last few years, which I think is a reflection of what’s going on in patent law. So with that our moderator today is, is always intimidating to introduce because like Cher and Madonna in the pat world, he only goes by one name. And you can say this anywhere in the country and everybody knows who you’re talking about.

 

There’s never a doubt. So I will–We have Yar from White & Case. Um, W– Yar now is I guess global head of White & Case’s IP group.

 

Used to be in the patent litigation ranks, now has emerged as both patent litigation and trade secret litigator. So with that, Yar, I will turn it over to you.

 

[YAR CHAIKOVSKY]

Thanks a lot. Appreciate it, Wayne. Um, re– really quickly, just to introduce, although others can say more, we’ve got Tom Spranklen from WilmerHale on my left then Rajiv Patel, partner at Fenwick & West, and then Camille Herdy from Rutgers Law on the panel.

 

If others want to make other comments of themselves, but won’t do that given we have our 45-minute clock that’s staring me in my face. Um, really quickly and, you know, obviously covering what Wayne just mentioned, I mean, we’re at the Advanced Patent Law Institute, and we’re gonna be talking about trade secrets, but it’s something that we’ve had to deal with in a post-AIA world and we’re migrating where technology’s going and AI, blah, blah, blah. I think one of the interesting table stakes for this is maybe some statistics with respect to trade secret cases. And with that, I’ll pass to Camille.

 

[CAMILLA HRDY]

Absolutely, yeah. So I think there’s a couple of reasons we’re here.

 

One is as you all know, right, there is now a federal trade secret law, De– Defend Trade Secrets Act passed in 2016. Also, there have been a lot of you know, potential weaknesses that we’ve been talking about today– Mm-hmm. In patent eligibility and enforceability.

 

And then third is probably because of AI. The trade secret AI interface is very interesting. And since I’m allowed to say one more thing about it myself, if you are interested in AI and trade secrets, you can download–I have several articles on this SSRN and then HRDY, and you’ll find those articles. But in terms of statistics, so since the DTSA was passed, the average number of federal trade secret cases brought per year is around 1,400, 1,400 a year.

 

There was an initial rise in cases right after the DTSA was passed just ’cause you suddenly had this automatic federal forum, but it’s now leveled off around 1,400 a year. Very interestingly, the win rate for cases that go to trial is very high, okay? So most trade secret cases do not go to trial, right?

 

They settle, you have an early injunction, and the case kind of just goes away. But when they do go to trial, it’s over 80% win rate for the plaintiffs, okay? That’s very significant.

 

Um, and then the third statistic I’ll just put out there is damages. Damages awards have been getting higher. They are so the median, so sort of like what most cases look like, is actually under a million, but the average is over eight million, and that is because you have some very, very, very large trade secret verdicts coming out that rival patent cases. So stop there.

 

[DAN ZEUNERT]

Okay. Thank you. I mean, a little bit of table setting.

 

And then next wanted to hit kind of a topic of defining what is or is not a trade secret in the era of AI in particularly, and maybe I’ll start with Rajiv in terms of, you know, things such as, I mean, and you can hit on what you want to, you know, whether AI generated c– you know, confidential information rises to trade secrets economic value for it, things of that nature.

 

[RAJIV PATEL]

Yeah, I mean, I think the primary thing that we’ll see oftentimes these days when we’re talking with companies about, “Hey, my AI created this trade secret,” well, if you think about how a lot of large language models work, by definition, they’re actually creating new materials, right? They’re taking in materials from elsewhere and creating new materials, and companies are saying, “That should be my trade secret in this context.” Unlike patent and copyright where I need a human to be in, basically the author for copyrights or the inventor for patent, trademark has no requirement of a human actually producing it in that context.

 

And trade secrets has no requirement for the human actually producing it. So companies are beginning to take a view where, “Hey, if I’m training my model to do something, and it spits out something new, that should be my trade secret that I keep,” right?

 

Whether it’s a new piece of code, et cetera, that goes along with it. Now, of course, that alone, the fact that it’s creating something that’s new that didn’t exist before doesn’t necessarily mean it’s a trade secret. As Yar just pointed out, it has to have those other elements that we require in a trade secret, such as economic value.

 

And so that has to be another element that we have to prove, and then same thing, you’ve got to also take a look at whether it’s something that’s readily ascertainable in terms of is what was created, something that, hey, it’s generally known or not in that context. And that’s actually where it starts to get interesting because it’s that what does that mean to be readily ascertainable? Especially with AI, when I can now use another AI to go reverse engineer what just went on, potentially and figure out– is that something that’s new or not?

 

Because here’s my spitting out of what I think is what you just created as the new AI.

 

[CAMILLA HRDY]

Yeah, great. No, I totally agree. It’s so interesting, and I love that Rajiv brought up the point about the fact that you don’t need a human creator, right? And that’s gonna be very significant because we’ve seen in the patent copyright space, no, you gotta have a human inventor.

 

You don’t for trade secrets, but there are a lot of limitations here. One, we were just talking briefly about ownership and who has standing to sue. We’re not gonna get too deep into that, but I will say DTSA says, the trade secret claimant has to be the owner, okay? State standards tends to be lax, or a possessor can usually sue.

 

Um, but the bigger barrier is what Rajiv was mentioning is, is you do have to show that your inf– you don’t have to show it’s novel and not obvious, Okay? So, we don’t have to use those words in this panel, but you do have to show the information is not generally known and not readily ascertainable to others through proper means and that it derives economic value from secrecy.

 

And so at the federal level, you do have to show information isn’t readily ascertainable, and in most states, you have to show that, notably not in California, but a lot of other doctrines perform the same function. Long story short, if another AI model can easily ascertain what the information is, may not be your trade secret, but we’re still working through this stuff, and these cases are only just coming out.

 

[TOM SPRINKLING]

And if I could jump in just for a second. So you’ve heard from the academic, you’ve heard from the transactional guy, I’m the litigator on the panel, and I am constantly faced with a barrage of different cases, different rules across jurisdictions.

 

So as Camilla mentioned, you know, one thing to think about here is that for many, many years, we had something called the Uniform Trade Secret Act, which most states largely followed, with some important exceptions. Then, about a decade ago, we switched to the Defend Trade Secrets Act.

 

Again, very similar with some distinctions. But when you start to look at the cases, a lot of different federal courts will draw from the state law precedent, even while there’s a whole new body of federal law emerging under the DTSA.

 

It can get very confusing very quickly. One thing we’re gonna talk about, I think, a little bit later is how you determine what you have to plead, for example, for a trade secret.

 

And that is a very hot issue in the Ninth Circuit, among other places. But I think some of these issues with AI in other areas are very confusing, very complicated because we don’t have a court like the Federal Circuit where these issues go up to, and they give a decision, and then that’s the answer. We have a bunch of different circuits, we have a bunch of different state courts, and so not only are these issues new and novel, but you won’t necessarily get one answer from the same jurisdiction.

 

[RAJIV PATEL]

Yeah. I think one area that’s getting interesting is, and it crosses over with that litigation is what does it mean to be readily ascertainable in this AI world, right? Because nowadays you can plug in stuff into AI and try to figure out what is going on.

 

So could you get the Coca-Cola trade secret formula? Theoretically, you could. But is that going to be readily ascertainable? Don’t know.

 

[CAMILLA HRDY]

Right.

 

[YAR CHAIKOVSKY]

I think you want to comment like on CompuLife case, etc. And that’s implemented there in terms of improper means/readily ascertainable.

 

[CAMILLA HRDY]

Yeah.

 

[RAJIV PATEL]

And when being readily ascertainable, one of the things is, and Camilla mentioned this earlier, just shortly, which is one of the things that some of the courts have done, and CompuLife did this, was by a human, right? And so, that becomes a critical element. So if you’re not familiar with that case, there was a case in which someone was going out and scraping a website relative to all insurance quotes to put together, kind of their offering and the like. But it was that scraping part that the court took issue with relative to whether or not that should be something that’s readily ascertainable.

 

The data’s out there. If a human wanted to, they could have just written down all the quotes that were right on that site. But because they used a bot to get all that data, court didn’t like that and didn’t consider that to be readily ascertainable.

 

[TOM SPRINKLING]

Well, actually, can I make one slight correction on this? So and maybe Camilla was about to jump in on this. I think it was potentially improper.

 

[RAJIV PATEL]

Oh, potentially improper, right. They haven’t quite ruled it yet.

 

[TOM SPRINKLING]

They didn’t address readily ascertainable question.

 

[RAJIV PATEL]

Didn’t go the full route, but it was leaning in that direction.

 

[CAMILLA HRDY]

Well, so but then follow on they did, so eventually.

 

[RAJIV PATEL]

In the second case.

 

[CAMILLA HRDY]

Yeah. So basically, the plaintiff won, right? And there are two issues, right? One is it readily ascertainable? Is it a trade secret at all?

 

Court very clearly says in that case that just because something is publicly available does not necessarily mean it’s readily ascertainable through proper means, right? So here, the way the defendants got it, they did a variety of naughty things, but the main naughty thing they did was hire a woman, the court called a hacker to design a bot to automatically get insurance quotes from the website. And so the claimed trade secret wasn’t just the individual quotes, it was the entire combination, this huge dataset, and the court says, “Look, a human could have gotten individual quotes easily through the site and largely for free, but they used a bot.”

 

So that was bad. But the second issue, and so not readily ascertainable by proper means, right? And I think they effectively did hold on that because we did get a holding that there were trade secrets and that they were misappropriated.

 

The second issue, it’s absolutely closely related, is have you used improper means? And you have to show both. You have to show you have a trade secret that’s not readily ascertainable through proper means, and you have to show the defendant used improper means to get the information. And this case, which I recommend anyone reading, it’s really one of the most important trade secret cases since DuPont versus Christopher which was that, you probably know this very famous case where two brothers fly a plane over a plant to see what’s inside.

 

They’re in public airspace, right? They’re not breaking any laws. Court says that’s improper means. That’s essentially what happens in CompuLife, right?

 

It wasn’t clear that they were violating any other laws, and then correct me if I’m wrong, but they lost on their copyright claim, right? So they won on the trade secret claim and that’s because we have this kind of roving concepts of it just looks bad, it’s improper, right?

 

[TOM SPRINKLING]

But just to pause a little bit though on this. Yeah, yeah. I mean, just going back to that theme about who knows kind of, it will depend heavily on the jurisdiction. Compulife is an Eleventh Circuit case.

 

[CAMILLA HRDY]

Yeah.

 

[RAJIV PATEL]

Yeah.

 

[TOM SPRINKLING]

And of course it doesn’t govern anywhere else.

 

[CAMILLA HRDY]

Absolutely.

 

[TOM SPRINKLING]

And so I do think, and it’s foundational, it’s cited in a lot of places, but I do think a question you might have going forward is what will other courts do in this circumstance–

 

[CAMILLA HRDY]

Absolutely.

 

[TOM SPRINKLING]

–when faced with something like AI? So one case that came to mind is the Kyllo case, which we all read in law school about the Fourth Amendment, and you might all remember this. This is the police use kind of heat wave detectors on the house and that’s deemed a search and it’s deemed a search because that’s kind of technology only the police had and not something the public would have.

 

I wonder whether that same kind of methodology might be used in other cases. It wasn’t used in CompuLife, but if you start thinking about kind of what are the reasonable expectations of someone–

 

[CAMILLA HRDY]

Yeah. Have they changed in 2025 versus 2020 versus 1995?

 

[YAR CHAIKOVSKY]

Well and even 2030.

 

[TOM SPRINKLING]

And 2030, sure.

 

[YAR CHAIKOVSKY]

I mean are you know the bots you know being improper at this point in time there’s improper. I mean, that’s the problem with the law in this area.

 

I mean, they’re just doing it more efficiently and more effectively. Humans could have done it. It would have taken a lot longer, or you would have had to employ, you know, an insurmountable amount of humans.

 

You know, it’s interesting. One of my partners I think is here. We’re on the appeal of the, you know, you know from a copyright perspective, it’s not a trade secret case, but we’re on the, you know, Ross Thomson Reuters case. And the reality is, when you’re amalgamating these headnotes that Thomson and Reuters found so find so precious, you know, you know, a human can aggregate that without a problem.

 

But the fact that, you know, you put it into a computer and the computer’s using it to train an LLM effectively, not for a Gen AI output, oh my God, you know, all of a sudden, because you’re doing it faster. But I could go to the law library here at Stanford and get those exact same head notes in the same way and they’re publicly available.

 

[CAMILLA HRDY]

Yeah.

 

[RAJIV PATEL]

I mean, the other thing that’s interesting, it is, I agree with you on the left, Eleventh Circuit’s case, it’ll be interesting to see how other courts look to that or not going forward. But arguably, they didn’t use a lot of reasonable measures to even try to keep all that information a secret.

 

And so, why couldn’t they go do that? There was no terms of service, for example.

 

[CAMILLA HRDY]

There was not. That’s exactly right. They did not initially have a terms of use that prohibited that activity.

 

They put one in after the case started, obviously. But they didn’t initially have that, and it’s worth noting that the terms of use does matter in these cases because courts will often look to was it prohibited by the terms of use in deciding whether it was improper or not, and you know, looking forward to 2030 or even to 2025 if you think about what’s the next step of this, right?

 

It’s trying to hack into AI models like ChatGPT and so forth, and we’ve already had some cases, or really one line of cases about that. But I think copyright life is gonna be very relevant here.

 

[TOM SPRINKLING]

Sorry, I can’t see you while you’re talking. There are layers upon layers of that, too, right?

 

Because one thing you might ask is, “Well, what type of AI are you using,” right? So, if there’s a publicly available type of AI and it’s very easy to locate this information in the compilation, that to me strikes me as maybe a slightly easier case for readily ascertainable. What if you have a proprietary AI system?

 

What if you come up with a thousand-word prompt that only you come up with and nobody else can possibly think of? So, I still think there are gradations in this as the technology advances.

 

[RAJIV PATEL]

Clearly, it’s fact-specific.

 

[TOM SPRINKLING]

Yeah, yeah.

 

[RAJIV PATEL]

Absolutely, it’s fact-specific, and I think when you’re thinking about like how you’re, at least with the way we talk about it with our companies when we sit down with them is you got to do as many steps as you can to help take those reasonable measures, and then to help with those gradation arguments later on. But if you don’t have any of that, then you could run into a case where you just don’t know where it’s gonna open up into, given that it’s still forming as law.

 

[YAR CHAIKOVSKY]

How about we migrate onto assertion of trade secrets and more particularly, let’s say, identification of trade secrets, whether it’s a pleading stage, perhaps later. I’m gonna have Tom kick that off and–

 

[TOM SPRINKLING]

Sure.

 

[YAR CHAIKOVSKY]

–get into our various standards, whether DTSA, California or otherwise.

 

[TOM SPRINKLING]

You’re music to my ears. So as, I, I will be, I’m only speaking for myself here. I generally tend to represent defendants, so I will try to take this from a neutral perspective, but I have a little bit of that in the back of my head. So California, which is of course where we are right now has what’s called the California Uniform Trade Secret Act, and it has a requirement that before discovery begins, you have to identify your trade secrets with reasonable particularity.

 

Now, what does that mean? If any of you can come up with a definition of that, I would love to hear it, but the basic idea, the basic policy concerns is that you should, the defendant should know what the trade secrets are before the case begins. Otherwise, if you start discovery, you can go rummaging around.

 

It could be a fishing expedition. So that’s great.

 

That’s fairly well-established law in California. Now, the Ninth Circuit has started getting involved in terms of what this means in the context of the DTSA, which again, very similarly worded to the California law, and there’s been a succession of cases.

 

There was a case in 2020 called Eteleclear, and there was a very recent case called Quintara, and Quintara holds that the DTA– DTSA standard is different. Don’t fully understand how it’s different.

 

I’m sure the Ninth Circuit will tell us too.

 

[YAR CHAIKOVSKY]

Well, just to clear, California– Petition of particularity, Right?

 

[TOM SPRINKLING]

So, it’s, they use the term, I’m monopolizing this very interesting topic called–

 

[CAMILLA HRDY]

Well, Just to be clear, there’s the California’s–

 

[TOM SPRINKLING]

Yes.

 

[CAMILLA HRDY]

Statute that says–

 

[TOM SPRINKLING]

There’s the California statute– Which says reasonable particularity, and there’s a fascinating case, which I commend to you all called Quintara from a couple months ago from Judge Johnstone on the Ninth Circuit, which says the DTSA applies a different standard, which is sufficient particularity

 

[CAMILLA HRDY]

Yeah.

 

[TOM SPRINKLING]

–which side note, seems to be drawn from California case law, but I won’t–

 

[CAMILLA HRDY]

Mmm, I think it’s drawn from–

 

[TOM SPRINKLING]

We could disagree. It’s drawn from INAX.

 

It’s drawn from INAX. In any event, I will stop talking there only to say that this is a front-line issue that we confront in all these cases. An early question in a lot of trade secret cases is, can you file an early motion to dismiss because nobody knows what the heck the trade secrets are?

 

This most recent decision from the Ninth Circuit would suggest it needs to wait till the summary judgment stage after there’s been discovery, after there’s been different details added. I will now pass over to other people who have, I’m sure–

 

[CAMILLA HRDY]

Oh, no. I couldn’t agree More, I mean. It’s actually a very big practical issue.

 

To me, it’s not like the most interesting issue in trade secret law, but it’s very, very important to trade secret lawyers because you do need to know, like, how do you file your case without getting it dismissed right away? And I think it really has to do with timing.

 

That’s my interpretation of all of this is that in California, because of the statutory reasonable particularity rule, which we all understand reasonable particularity, sufficient parti– particularity. Those words themselves mean nothing, but the statute does say that you have to identify your trade secrets early in the case, before you enter discovery, right? And the policy there is so that the other side knows what they’re being accused of and so that we don’t go, on a fishing expedition and discovery balloons, et cetera. But the DTSA doesn’t say anything about that, right?

 

It does not have that reasonable particularity language or any, it doesn’t mention identification at all. And so, I think, I actually think the outcome in this case is right. I think that the court is saying, “Look, California has a statutory rule, says you need to do identification very, very early.”

 

The DTSA doesn’t have that statutory rule, nor do many, many, many, many states where the standard is just laxer. And it’s not that you don’t have to identify your trade secrets, it’s that you can do it later. I think the court might exaggerate by saying, you know, this usually isn’t gonna be figured. It’s a question of fact, usually won’t be decided until summary judgment or trial.

 

I think that might be an exaggeration, but still you can do it later, so what I’m telling my students, ’cause I have to teach this stuff, is like, under the federal– Quintara suggests, and that’s just the Ninth Circuit, but they’re suggesting that the DTSA is a bit laxer than California. But frankly, that’s been kind of the trend in a lot of other cases, too.

 

You will see other circuits that are also not necessarily saying, you have to identify everything right away before trial–

 

[TOM SPRINKLING]

It’s an interesting–

 

[CAMILLA HRDY]

–or before discovery.

 

[TOM SPRINKLING]

It’s an interesting question. I might push back on the question of whether it’s rightly decided or not, but I will defer that for this panel.

 

But what I would say is you’re right. It’s coming up in a lot of different circuits across the country, and there are several interesting more academic issues that come up with that. For example, is it a question of law, right?

 

Quintara holds it’s a question of fact, how detailed the trade secrets are, but judges look at this kind of stuff all the time, right? And they think about like, “What is the scope of the patent right?

 

Why should this also be a question of law?” That’s something that I think has not been fully settled.

 

[CAMILLA HRDY]

That’s a good point,

 

[YAR CHAIKOVSKY]

Can I ask a question, just ’cause the one thing I also disagree with is how many people here are in-house, actually at a company? Okay, right.

 

And why you should care, obviously, and why this issue is top of mind is expense. I mean, you know, it actually is one of the most significant issues in trade secret litigation that we deal with, because as opposed to a patent case, which has its metes and bounds with the claim, and we kinda know what we’re dealing with, subject to Markman, which is kind of, you know, we can debate what Markman’s become today. But you know, here we have just this obtuse IP right that is ill-defined and when are we gonna get it defined?

 

And we got a court saying, “Oh, later,” right? Where the California law is, well, there’s no discovery until it’s defined sufficiently, right?

 

[TOM SPRINKLING]

I’m happy to take the California position.

 

[YAR CHAIKOVSKY]

Right. And so, you know, here the worst thing is the Ninth Circuit saying, “Not California. Different later.” And so, from in-house and especially if you’ve been involved in trade secret litigation, the costs, I mean–

 

[RAJIV PATEL]

Go up.

 

[YAR CHAIKOVSKY]

–necessarily rise, I mean, I’m not gonna say exponentially, but significantly. You know, and the cost of trade secret litigation is we just heard the stats at the outset. We’re like, “Hey, we’re gonna have like 80% of these things head to trial.” You know, you gotta get this reduced somehow, and one of the methods we use to reduce this is this identification and trying to, you know, cut the case down.

 

And the case law, at least on the DTSA side, if you’re on the defense side, it’s not going the right way for early identification, et cetera. And so if you’re a plaintiff, you love it.

 

[RAJIV PATEL]

Yeah, absolutely. I mean, it’s not gonna be just, what you gave were the damages awards, just to get to even through discovery is gonna easily cost you a million bucks just to try to figure out what is the trade secret you’re arguing over.

 

[YAR CHAIKOVSKY]

More.

 

[RAJIV PATEL]

Yeah, probably more, right? At least a million is what I’m getting at, right? It’s killing the companies that are trying to figure out like what is there. Plus on top of that, you also have the, there’s only two states, I think, that have the reasonable particularity right?

 

Is California, and Massachusetts, and arguably, South Carolina, maybe. So it’s gonna be interesting to see if more states go in the federal route, right? Which is gonna mean less certainty and definitely gonna be more impactful as to the budgets.

 

[CAMILLA HRDY]

Yeah. I guess I would just say that just because you’re saying California’s rule doesn’t apply to federal law or everywhere, that doesn’t mean that all cases are gonna go to trial. I think whatever sufficient particularity is, I do think it means something. We do still have an identification standard and if you look at cases like Oakwood and so forth from other circuits, I do think it’s possible to kick out cases early. I just think the California standard, Why would the California standard apply to the DTSA?

 

That part of the opinion has to be correct.

 

[TOM SPRINKLING]

I would raise similar questions about why InteliClear treated it. But you know, I mean, if we’re thinking to defend Judge Schonstone and his decision for a second, I think the other thing that jumps out to you is he suggests that maybe you could have a very quick summary judgment decision, which it doesn’t save you the problem of getting to motion to dismiss. I recognize that, but the suggestion from the Ninth Circuit is you tee this up very, very early in a summary judgment posture rather than in a motion to dismiss posture.

 

It seems to be the way they’re heading.

 

[YAR CHAIKOVSKY]

Yeah, yeah.

 

[YURI ZHUKOVSKY]

That’s, I was just in a courtroom in New York and applying Second Circuit law with sufficient particularity and someone’s, we’re actually the plaintiff in that case and, you know, the judge is struggling. He’s struggling with what is sufficient particularity. I mean, it’s part of also the problem is there’s not enough clarity. So if someone wants to bring this, whether it’s at the motion to dismiss or the summary judgment stage, depending on the trade secret, the type of trade secret, I mean, here we’re talking about AI code, actually relating to mass transit.

 

You know, it’s very hard for the court to discern what is sufficient particularity, and so then you’re gonna get into fights about it, fights about when, and all of a sudden, you’re fighting all the way to the end of discovery. This is where, again, I’ll disagree with our professor. You’re fighting to the end of discovery if you’re on the defense side.

 

If you’re fighting to the end of discovery, you’re more in trouble. Sure, you’re hopeful at summary judgment stage, but obviously, we want to try to pare this down before we get to that point in time.

 

[RAJIV PATEL]

Well, you also have to worry about, I guess, and then I haven’t seen it, but isn’t, I don’t know if anyone on this panel has, statute of limitation is only three years, and if you don’t define what it is in that context, both for the DTSA and also California, do you run the risk of losing your case if you’re a plaintiff?

 

[CAMILLA HRDY]

Yeah, yeah. We were gonna talk about the statute of limitations as well. Should we?

 

[YURI ZHUKOVSKY]

Yeah, you got to cover it right now. Sure. Yeah, let’s go.

 

[CAMILLA HRDY]

Okay, yeah. So another interesting issue is how quickly the plaintiff has to sue, right? And so the DTSA has a three-year statute of limitations, but it has a discovery rule, so it’s three years after the plaintiff discovered the misappropriation or could using reasonable diligence have discovered it, right?

 

So basic discovery rule. And the case I really like, I think other people wanted, maybe wanted to talk about a different one. The one I was gonna talk about was Insulet versus Eoflow, which is pretty interesting on this.

 

This case did go to trial, plaintiffs prevailed, and this, these were insulant patch pumps, so important product competing product case. Insulet’s the plaintiff, some of Insulet’s employees go to EoFlow. Eoflow comes out with a competing insulant patch pump, looks pretty similar.

 

And the plaintiff did sue but they sued more than three years after, according to the defendant, they could have learned about the misappropriation. And the thing that the defendant said plaintiff knew and that put it on notice was that the defendant presented a prototype of its product at an industry conference, okay? And so the idea says the defendant will combined with the fact that, you know, your employees now work for us, you saw this prototype at a conference, you should have sued within three years after that, but they waited a little longer. The jury found for the plaintiff anyway.

 

The court upholds that and says, you know, “Mere suspicion isn’t enough, seeing a prototype from ‘afar’ shouldn’t necessarily tell you it’s misappropriated, you know, trade secrets.” And in fact, I think there was kind of a good faith notion here, like, “We’re gonna take them at face value, they said they’ve independently developed this, they, you know, all companies should be given a little bit of good faith, and why should we necessarily think that it’s, with stolen IP?” So that tracks other cases I’ve seen, like something like defendant has asked plaintiff for a license and plaintiff says, ‘No, you can’t have a license.’ Does that put you on notice that there’s gonna be misappropriation? No. Like, we should kind of assume good faith, I think is kind of the, but, yeah, Tom?

 

[TOM SPRINKLING]

Well, I was just gonna say one quick thing about Insulet, though, is, so first of all if you’re a plaintiff, file your case early. Don’t wait, you know, I mean, gather facts, but be conservative on this because it’s a shifting area of the law, and I would never suggest otherwise. That said, I do think the one notable thing about Insulet, is that the rule it announces, it has a pretty exceptional paragraph where it says, “I recognize no other court interpreting the DTSA has come to this interpretation of the statute of limitations.”

 

That is, something other than inquiry notice. And it applies the notice, a standard it takes from a Supreme Court case in a different context called Merck, I think. And I actually don’t think I fully understand some of what the court says could be read in its analysis section, I think is something close to inquiry notice, because it doesn’t grant summary judgment for either party.

 

It says, well, you know, there’s some facts one way, some facts the other way. But I do think this is a place where courts are split and still trying to figure out exactly what the rule should be.

 

[CAMILLA HRDY]

Absolutely.

 

[TOM SPRINKLING]

I will say again, in California it is, it is the diligence test, it’s the higher standard so that’s pretty well covered. But on the DTSA, I’m not certain.

 

[YAR CHAIKOVSKY]

But you say higher standard and then you should have known standard, et cetera, you know, what would you say with respect to that prototype?

 

[TOM SPRINKLING]

Well, as a defense lawyer, I would say obviously they should have known and therefore the statute of limitations has run. On the basis of the facts, I think it’s a hard call.

 

[YAR CHAIKOVSKY]

Right.

 

[TOM SMEDINGHOFF]

And I think a lot of these statute of limitations questions are gonna be very, very fact-intensive, and it may depend on who has the burden of proof. Usually, it’s an affirmative defense. In California, at some point, it switches to the plaintiff if you make certain showings.

 

[CAMILLA HRDY]

Would you agree that if there had been downloading activity discovered, The fact that we got this prototype, looks similar from afar. We’ve got the fact that their former employee is now with the defendant, would some kind of suspicious downloading activity, would that clinch it?

 

And then you should discover it.

 

[TOM SPRINKLING]

I’m unwilling to make an advisory opinion at this time.

 

(Camilla Hrdy laughing)

 

But I certainly think it would be a closer case. It’s helpful. It’s helpful.

 

[YAR CHAIKOVSKY]

Is there anything we do transactionally in this area?

 

[RAJIV PATEL]

Yeah, I mean, I think the thing that’s screamed for me was just like the layering that you need to do relative to the IP you should be doing. I think trade secret cases alone, you’re just hearing. A lot of unknowns in them, right? And so, in this case, being able to say like, ” Oh, in addition to that product, we’ve got a patent on it, we’ve got copyrights in the code that’s within it,” et cetera.

 

Whatever you’re doing, layering these cases is probably gonna be more helpful if you’re on the asserting side, the plaintiff’s side, that is. And as a defendant, you know, in some cases, that’s gonna make you a little bit more nervous, too, because now you’re not just dealing with all of the issues involving the costs of the trade secret-related issues, but now you’ve gotten these other cases that also are, we all know patent cases are expensive, and now copyright cases are getting expensive too in that context.

 

So, I think layering all of this together as part of the early stage as a strategy, you need to be thinking about as a company for a lot of the in-house folks here.

 

[TOM SPRINKLING]

The other thing we haven’t spoken about is that an ongoing misappropriation kind of counts as one misappropriation. So I suppose if you had a purported misappropriation that started in the year 2010 and continued all the way up to 2024, you wouldn’t be fighting with these difficult issues.

 

So that’s also worth considering as well, if it’s an ongoing course of conduct.

 

[RAJIV PATEL]

Yeah.

 

[YAR CHAIKOVSKY]

Anything else that you do on the statute of limitations from a contractual point? You know, is there anything like an owner of trade secrets could do to kind of make it more readily apparent that the employee’s taken off, let’s say, developed something that’s got a prototype that’s substantially similar?

 

Anything?

 

[RAJIV PATEL]

I mean, clearly your employment agreements are gonna articulate in there that you are the company, your employee is the owner of those trade secret. Helps you from the perspective of the DTSA because they define these trade secrets in the context of the owner.

 

Doesn’t necessarily help you from California state law or other Uniform Trade Secret Law states out there because they talk about things in the context of possession of the trade secret, and that’s not necessarily the same as being the owner of the trade secret. So it gets a little bit trickier on the state side relative to these issues, because there you just, as the person who developed the trade secret, if you can argue that I was the one who possessed those, you’ve got rights under California Trade Secret Act, whereas if you go to federal, the DTSA, you gotta show you’re the owner of it. So there’s a little bit of a so in the contracts, while you can have it as the owner, you still have to worry about the possession side of it too.

 

[TOM ZICKELBERG]

And more than one person can possess or own a trade secret, right? If it’s independently developed, 10 people could possess it as long as they meet all the other standards. So it’s a little more fungible in some ways, and something like that happen–

 

[CAMILLA HRDY]

I do think it’s worth mentioning the really interesting ownership case. It, it goes in the weeds, but it is a Snyder versus Beam 2025 Tenth Circuit case, where the fact pattern was that an employee who had worked for another company, Guardian, goes to his new employer, uh, Beam Technologies, and he brings with him from his former employer trade secrets, a client list from his former employer.

 

And he has it at his new employer, and he accidentally distributes it to his new employer, then he gets into a fight with them, too, and he leaves, he sues the second employer, they misappropriated his trade secrets ’cause they obviously took the client list, he had very generously brought. And you would think, “Well, that’s ridiculous. How can you sue on a client list that you took from your prior employer?” To me, that’s just, like, patently ridiculous.

 

[ERIC CHAIKOVSKY]

Yeah.

 

[CAMILLA HRDY]

But that is where the law has gotten us, because yes, DTSA says you have to be an owner. State laws don’t say that for almost all the UTSA versions in the states.

 

They talk about possessor or complainant, person. They do not say you have to be an owner. And so we’ve gotten this ridiculous rule that a mere possessor can potentially sue for trade secret. So I could steal Rajiv’s trade secret and then sue you after you take my trade secret.

 

I think it’s ridiculous But you know, that’s kind of where we are with the state standing rules, is mere possession can be enough, even if it’s derived. And Snyder versus Beam seems to support that. It says under the DTSA you got to be an owner, but under Colorado trade secret law, which this case was under, this guy might have had standing to sue. He still lost the case ’cause, you know, obviously he was gonna lose, but he didn’t take reasonable measures.

 

He distributed very freely to his current employer.

 

[RAJIV PATEL]

This goes like, I mean, every type thing you read about, right, is like when you’re at a company and you signed that agreement, employment agreement, and the like, it says like, “Oh, this is all company property,” et cetera. And then customer lists and contact lists are always included as being like, of course that’s company trade secret.

 

And yet you’ve got this–

 

[CAMILLA HRDY]

And so to be clear one of the really weird things was that the original guardian entity wasn’t in the case. If they’d been there, if they’d had the ability to speak up and say, “No, no, no, no. We own this.”

 

Then it would just be a basic employee ownership case, and obviously, they would say, “We own this. It was developed on our

 

[RAJIV PATEL]

on our time. on our time.

 

[CAMILLA HRDY]

And this also, but they weren’t there.

 

[TOM SPRINKLING]

This also makes an interesting point, which is I’m, this is a sophisticated audience, but um, sometimes when, when folks say the words trade secrets, they think something that’s like a patent but isn’t quite patentable. So they think it’s something kind of AI-ish, a new cool machine, or something like that.

 

It could be a client list. It could just be three names written down. It could be a formula for Coca-Cola or oil. It spans all kinds of IP rights.

 

And one reason why it’s very interesting to do, some of these cases end up looking more like kind of like classic business cases, you might have torts involved. Some of these cases are patent cases, they have the tech side added onto it. So it can kind of go all over the map.

 

And I think sometimes people get into trouble because they don’t think, maybe this is a trade secret before sharing it.

 

[CAMILLA HRDY]

Yeah, and it’s worth noting on the stats over recent data suggests over 40% of the federal trade secret cases do involve business relationships like client lists and so forth. It’s very common.

 

[TOM SPRINKLING]

Yeah.

 

[CAMILLA HRDY]

Not as interesting to IP lawyers, probably.

 

[TOM SPRINKLING]

No, and speaking of do we want to talk about damages next?

 

[YAR CHAIKOVSKY]

We, yeah.

 

[RAJIV PATEL]

I think– Well maybe

 

[TOM SPRINKLING]

Before we get to those, unless you got, you got anything else– One more thing, I was going to say is, and then let’s pivot to damages.

 

[RAJIV PATEL]

Interestingly, though with AI now, there’s now a lot of discussion about, well, I can recreate that list using these AI models.

 

[CAMILLA HRDY]

Exactly.

 

[RAJIV PATEL]

Et cetera, and so should it be a trade secret anymore? Are we back to having to define whether or not that should even be a trade secret because it’s going to be readily ascertainable using a lot of the AI systems that are out there, and may not rise to the level of saying, well, that’s beyond human capabilities or whatnot. Because there’s so much more information out there, and AI is just helping me organize it. That I’m recreating it.

 

[YAR CHAIKOVSKY]

Okay, you want to start damages, Camilla?

 

[TOM SPRINKLING]

Uh, well, all I was going to say and I, I think others may have more to say on this than I do, is one reason why damages are on the rise in trade secret cases, and I think Camilla has statistics on that she’s already given, is because they’re often very close to things that are in the patent context, but maybe they’re not patented because the information the owner doesn’t want to share the information publicly, or maybe it falls under section 101, it’s too close to mathematical formula or something like that. But the amount of money at stake in these cases now rivals or dwarfs patent cases. And that’s also very notable because the case law is very underdeveloped. So I want to just, one thing.

 

And then we’ll pass it off to the rest of the panel. As you all know, in patent law there’s apportionment, right? The patent owner must show the value of the infringing the patent in the infringing product. In trade secret law, it’s very mixed about whether there is an apportionment requirement, who has the burden on it.

 

Some people have gone so far as to argue it’s enough to say your product has my trade secret in it. Now you need to tell the court how much of it is not attributable to the trade secret.

 

So you have very high dollar amounts, and you have a lot less defenses for a defendant in this context.

 

[YAR CHAIKOVSKY]

Just really quickly, it, I mean, we’re all doing patent litigation, everybody doing it. I mean, obviously, we’ve seen over the last whatever it is now decade in terms of the Federal Circuit’s pairing of damages awards. We kind of count on the large damage awards being reduced for one reason or another, being sent back down, remanded, remitted, et cetera. You, we don’t have this in trade secret law.

 

We may one day. But we don’t, right? We have the various circuits around in the United States, and given what Tom just mentioned in terms of what people are doing as, as people do apply, for example apportionment and kind of reasonable royalty analysis, but you also then have the big unjust enrichment component.

 

And the bigger numbers tend to come from the unjust enrichment component. And that one’s a lot harder to let’s call it fight back and the appellate court’s going to be dealing with that, these large numbers coming out of these cases under unjustement, unjust enrichment, among other theories.

 

And oh, by the way, what you’re starting to see in these cases with DTSA or RICO claims, you know I’ve been doing that on both sides because you’ve got conspirators and all of a sudden, you’ve got treble damages, you are in the mix automatically so you’ve got a DTSA with a RICO claim, and it’s all of a sudden off and running. So you know, numbers are big here so beyond the expenses we already talked about, the ill-defined IP, et cetera, the numbers start to get bigger.

 

Which also then raises the stakes and the cost of litigation.

 

[RAJIV PATEL]

Although, and the, I mean one of the reasons numbers may be bigger, is there’s just look at Silicon Valley in general, right? There’s a lot of movement of really top-end talent in the AI space right now. And that movement is coming at a high cost relative to the people who are acquiring that movement, right?

 

[CAMILLA HRDY]

There are going to be more trade secret cases in the wake of these AI movements,

 

[RAJIV PATEL]

That are losing that talent are like I can’t allow this to happen. I’ve gotta do something here, and they’re gonna make these arguments and say, you know, so you can start with and saying, “Look at how much they paid this person to go over there.”

 

[CAMILLA HRDY]

And why are they paying them that much?

 

[RAJIV PATEL]

Exactly.

 

(Camilla laughing)

 

[TOM SPRINKLING]

And a trade secret, unlike a patent, doesn’t expire. All you have to do is keep it confidential. And that, I guess, all you have to do is understating the burden on the owner. But a patent has a limited life. A trade secret does not.

 

[CAMILLA HRDY]

Disagree, but anyway. Just kidding. I won’t get into that.

 

[TOM SPRINKLING]

Statutorily, there’s no life on a trade secret. We could discuss the doctrine later.

 

[YAR CHAIKOVSKY]

Anything else on damages? Any comments? Anybody else with something?

 

[CAMILLA HRDY]

No, I mean, didn’t fully, did you wanna do extra-territorial issues, or we could also talk about the open evidence case, it’s interesting, but also could be interesting to talk about sort of the TikTok and iTera cases.

 

[YAR CHAIKOVSKY]

Before we do that, to make sure I check the box here, I, know Rajiv’s done a little bit of this, it’d be better to get come from all given here. We’re at The Advanced Patent Law Institute.

 

Just comments, and then we’ll hit the international aspect, but comments on people evaluating, you know, and they, we’ve threaded this through this. But just overall, the difference between trade secrets, patents, copyrights, as you see it today, and maybe I’ll start with the person who’s doing it on the transactional side first.

 

[RAJIV PATEL]

Yeah, I mean, historically, we’ve all known that, okay, when we’re looking at patents and stuff, general conventional wisdom was there’s a way of detecting it. I’m gonna try to find a patent, let’s say. That’s the way what most companies thought about it, and if I couldn’t, I was gonna keep it a trade secret.

 

Not necessarily sure that’s necessarily holding as the line these days, and there’s a lot more blurring of that. And the reason is, is because of the way information is moving and flowing is a key reason.

 

So we had a situation where we had some code developed by a company, and that code was what they considered to be a trade secret, but it just so happened that they, company thought, “You know what? We should also just go ahead and file a patent on this. We won’t file internationally. Rather, we’ll just keep it, you know, through the patent process.”

 

So it’s gonna stay confidential for perhaps three to five years depending on how long the prosecution process takes. What happened during that period of time, though, was that an employee left and uploaded that code into a GitHub so that when they went to their new employee, they ended up downloading it there.

 

Now you would think like, “Why would you do that?” But this happens. We’ve probably all run across situations like that where common sense sometimes doesn’t necessarily take root that easily. What was interesting was that prosecution had already been started and it got accelerated, got the patent issued, and suddenly it was not just a trade secret misappropriate, well, patent wasn’t it was a trade secret misappropriation case that was brought at that point. But also, as soon as that patent issue is gonna be layered on with the patent lawsuit that was there and ended up settling much quicker because they, the other side just didn’t wanna deal with that aspect of it altogether.

 

But it’s just showing that a situation where, like, oh, historically no one would have ever filed a patent there, they just happened to do one there, but it got the company thinking, saying, “Should we be looking at this more critically than just saying, ‘Oh, I can’t detect it, we’re just gonna keep it a trade secret.'” Should we look at it more critically and not just, “If we can’t get a patent, should we be looking at things like perhaps I should register the code there?” Because remember when you register the code at the copyright office, there is provisions there that allow you to keep some of that code a trade secret, besides you’re not also downloading the entire code into the registration office when you’re registering that code. So you could also layer on the copyright claim there too if you ended up in that misappropriation situation.

 

[CAMILLA HRDY]

Assuming you had substantial similarities, yeah.

 

[RAJIV PATEL]

Yeah.

 

[CAMILLA HRDY]

I mean,

 

[TOM SPRINKLING]

One thing that you noted that I thought you might get into is this does happen sometimes, where people have a trade secret. And then they file for a patent application, and they put the trade secret in the patent. You should not do that.

 

[RAJIV PATEL]

Do that, no.

 

[TOM SPRINKLING]

That’s just to be 100% clear. Especially now that we have published patent applications.

 

[RAJIV PATEL]

Yeah.

 

[TOM SPRINKLING]

I thought that’s where you were going,

 

[RAJIV PATEL]

Yeah. No. So functionality on the patent side, right? That’s what you’re covering there as opposed to the particularities of what is a trade secret, which would be some of the aspects of the lower level aspects of the code that’s in there.

 

But it’s that layered approach again that’s going in there. And same thing on the copyright, really. In the copyright, again, you’re gonna register it using those provisions relative to that, and yeah, well, it’s, that’s potentially if you’re just downloading the whole code, you’ve actually just copied it, right Yep,

 

[CAMILLA HRDY]

Yeah,

 

[RAJIV PATEL]

Yeah, yeah. Yeah. So it’s straight out there.

 

[TOM SPRINKLING]

And one other distinction I think that’s worth raising between patents and copyrights on the one hand and trade secrets on the other is, at least as of now, an AI could create a trade secret. And it would be a legally enforceable right. I mean, maybe there’s a district court case somewhere I haven’t read, but it’s certainly not something that’s been well explored.

 

There are cases now that say an AI cannot, you know– A patentable work or copyrightable work cannot be primarily derived from AI. I think, I think there’s a term of art I’m probably not using, but it has to be mostly human or sufficient human support. So it actually is a world where AI could do a lot more and create valuable resources.

 

[CAMILLA HRDY]

Yeah, yeah. And so that’s gonna put more pressure on trade secret

 

[TOM SPRINKLING]

Doctrine, exactly,

 

[CAMILLA HRDY]

Yeah. In terms of what can be protected, yeah. Very interesting.

 

[YAR CHAIKOVSKY]

Yeah. Happy to move on, you know, running a little short in time if people wanna hit the international aspects as you already mentioned, Camilla, in terms of you could start with the Itera–

 

[CAMILLA HRDY]

Oh, yeah. –idea?

 

Maybe just, I mean, not even going to the cases, but just say, you know, when you have a foreign defendant, it’s often gonna be the case that the DTSA may well apply because the DTSA applies if the, excuse me. The DTSA may not apply because what you need is a U.S. citizen or an act in furtherance in the U.S. But we’ve had a lot of cases, including iTera, but also recent TikTok case that settled, suggesting that the act in furtherance can actually be a wide variety of things.

 

So it could be something like poaching employees from a California, so say there’s a Chinese company who’s getting sued, but they poached employees from a California company. Okay, that might be enough. Also, even just not really having any U.S. activity, but then selling products after the fact, in the U.S., made with trade secrets that you got from those companies. That’s like barely minimum contacts in the personal jurisdiction Civ Pro world, but that seems to be enough in a lot of these cases.

 

So act in furtherance is something that’s developing. That’s kind of, I think, the big issue there is, are you gonna get your DTSA?

 

[TOM SPRINKLING]

The only thing I was gonna jump in to say is that you might all be wondering what happened to the presumption against extraterritoriality? And the answer is the DTSA, or rather the predecessor, kind of eviscerates it.

 

[CAMILLA HRDY]

It does. I mean, yeah.

 

I mean, and I think that’s correct. Like, you read the statute, and yeah, sure, it was hooked onto a criminal law passed in 1996, but it’s still the case that 18USC 1837 does say, right, if the, you have a U.S. citizen or act in furtherance in the U.S., this law will apply. And so I personally think the courts in those cases are right, that the presumption is overcome.

 

[TOM SPRINKLING]

Yeah. I mean, the only point I was gonna make is that it’s a statute that expressly says it, which I think is fairly unusual. Especially if you think about the sweeping nature of the rights that we’ve been talking about and the amount of money. I think what constitutes an act in furtherance is hotly debated and will be the subject of many interesting cases as we go forward. I think there’s, it’s the Hytera case, right, that says that an act in furtherance doesn’t even need to be a consummated act of misappropriation.

 

It could just be one tiny step, and what that requires is gonna be litigated, and maybe other courts will go the opposite direction and try to effectively, you know, enforce the presumption. I don’t know, but

 

[YAR CHAIKOVSKY]

Yeah, I mean, I was gonna go with Tom’s comment overall and for those that have been litigating trade secrets cases, the act in furtherance decisions, so far seem to be going more in the Hytera direction, you know? And now it’s playing it friendly, shall we say? But I would agree, we don’t have enough courts, let alone the Supreme Court commenting, despite there’s precedent for where that, the phrase came from. But applying it to the specifics of the case so that we have at issue, what’s at issue?

 

Because then what we’re talking about when you have these international cases, I don’t care whether they’re from China or any international aspect, I mean, it just allows for extraterritoriality in every trade, every DTSA case. And the damages number, though again, kind of constant theme outta my mouth. Whether it’s reasonable particularity, et cetera.

 

If you’re on the defense side of this and someone can actually make these arguments about act in furtherance, and that is the law, Seventh Circuit starting it, you know, we’re all of a sudden defending against multi-billion dollar claims. And gotta put more resources on it and more resources on defending the international damage calculations, and they’re asking the companies to then turn over their financials from across the globe. It just becomes a different escapade.

 

Uh, and I’m not sure that was what was intended when the DTSA was enacted is the delta.

 

(laughing)

 

[TOM SPRINKLING]

This just circulates, I mean the theme of this panel I think, has been if you have a trade secret question, there likely isn’t an easy answer. There’s gonna be different answers across different jurisdictions.

 

The law is still developing in nearly every area, and it could take a sharp left turn or a sharp right turn and we just don’t know. The Supreme Court has yet to take A DTSA case.

 

[CAMILLA HRDY]

And they denied cert in the CompuLife case, which–

 

[TOM SPRINKLING]

That’s true.

 

[CAMILLA HRDY]

To me was very frustrating.

 

[TOM SPRINKLING]

The CompuLife case or damages cases that have gone up, if you look at the Supreme Court’s precedent on trade secrets, it’s mostly a case called Kewanee Oil from the 1970s, which says that state trade secret laws are not preempted by federal patent law.

 

[CAMILLA HRDY]

Thanks.

 

[TOM SPRINKLING]

So there’s a lot of really interesting stuff that could happen, and so anything you hear, I think is just subject to further developments and arguments by academics and whatever.

 

[CAMILLA HRDY]

Yeah. And Josh, who has co-authored a what is basically a treatise in Understanding Trade Secret Law with his dad, and I can say it’s very, very good. I’m using it to prep for my class, Understanding Trade Secret Law by Josh.

 

[TOM SPRINKLING]

Well, I’m always, I should say, I’m replacing Josh Lerner on this panel, but I am Tom Sprinkling. My father is John Sprinkling.

 

[CAMILLA HRDY]

Oh, yes.

 

[TOM SPRINKLING]

He is written a very effective treatise that I co-authored, and he’s also actually written an article on trade secret in the artificial intelligence age, which was published in the South Carolina Line of Law Review last year, which I commend all of you and his thoughts are his own, but they’re pretty smart thoughts.

 

[CAMILLA HRDY]

But there are more recent articles on AI and trade secrets by me.

 

(laughing)

 

[TOM SPRINKLING]

Yes. No, yeah, for sure. For sure.

 

[RAJIV PATEL]

Do we have time for questions or no?

 

[IAN BALLON]

We can take questions during break. How about that?

 

[RAJIV PATEL]

Okay, we’ll know after. Thank you, everybody.

 

(audience applause)