Suppose the letter reaches a four-person sealant company in Duluth on a Tuesday. It comes from a much larger competitor's outside counsel, it runs six pages, and it says that Tessaly Sealants has "misappropriated Corvane's proprietary formula" for a low-temperature marine caulk. It demands that Tessaly stop selling the product within ten days, preserve all documents, and identify every person who has worked on it. Tessaly's head chemist spent eleven years at Corvane, left six years ago, and spent much of the time since then ruining batches in a converted garage until one finally cured properly at minus twenty. She read the letter twice and said what people in her position nearly always say: "I built that."

She may be right. Being right, though, is not yet a defense. A defense is a body of evidence, organized against the elements of a statute, and assembled early enough that it can still be trusted. The law gives her real room to work with. Trade secret law, unlike patent law, has never prohibited someone from arriving at the same answer honestly. Congress wrote that into the federal statute in so many words. But the honest answer has to be provable, and a former employee who produces a similar product is exactly the fact pattern from which courts allow juries to infer copying.

This article is about how those claims get defeated. MC Law has separate documents on identification (Identifying Trade Secrets in Litigation: Reasonable Particularity, Sequencing Discovery, and Protecting the Secret While You Sue), discovery (Managing Discovery in a Trade Secret Case: A Practical Guide), and money (Trade Secret Damages and Remedies: Unjust Enrichment, Reasonable Royalties, Injunctions, and Exemplary Awards). Here the question is narrower and more practical: when the accused party says "we did not take it," or "there was nothing to take," what does the law actually require, and what does the proof look like?

The short answer

  • The plaintiff carries the load on nearly everything. It must prove a specific secret, reasonable measures to protect it, value from secrecy, and acquisition, disclosure, or use by improper means. Independent development is usually a denial of use, not an affirmative defense, so the burden of proof stays with the plaintiff even though the defendant will have to come forward with evidence once access and similarity appear. Sargent Fletcher, Inc. v. Able Corp., 110 Cal. App. 4th 1658 (2003), is the leading explanation.
  • Reverse engineering and independent derivation are lawful by statute. The Defend Trade Secrets Act (DTSA) says improper means "does not include reverse engineering, independent derivation, or any other lawful means of acquisition." 18 U.S.C. § 1839(6)(B). The Supreme Court said the same thing about state law in Kewanee Oil Co. v. Bicron Corp., 416 U.S. 470 (1974).
  • Information that is generally known or published is not a secret, however it reached you. A patent or a published application ends secrecy for what it discloses, and a foreign patent counts. Attia v. Google LLC, 983 F.3d 420 (9th Cir. 2020); Ultimax Cement Mfg. Corp. v. CTS Cement Mfg. Corp., 587 F.3d 1339 (Fed. Cir. 2009). A public document nobody noticed is a harder argument than a patent.
  • "Readily ascertainable" works differently in California. Federal law and most state statutes make it part of the definition, so the plaintiff has to negate it. California removed it from the definition and treats it as a defense, which appears to require proof that the defendant actually got there that way. ABBA Rubber Co. v. Seaquist, 235 Cal. App. 3d 1 (1991).
  • Combinations and compilations are the trap. Showing that each piece is public does not end the case if the combination is the secret. Minnesota Mining & Manufacturing Co. v. Pribyl, 259 F.3d 587 (7th Cir. 2001).
  • Limitations can erase even an enormous verdict. In Insulet Corp. v. EOFlow, Co., No. 25-1807 (Fed. Cir. May 28, 2026), a divided panel reversed a judgment built on a nine-figure jury award because the plaintiff's own emails showed it had enough to sue more than three years earlier.
  • Good engineering records are the best evidence you will ever have, and they are not self-proving. A git history shows the path of development, but commit dates are set by the client machine. Corroborate them with records a third party holds.
  • The fee statute cuts both ways, and it is hard to win under. Bad-faith fee awards happen, sometimes in the millions, but losing on the merits does not establish bad faith. Most courts require both an objectively specious claim and subjective bad faith.

What the plaintiff has to prove, and why that is your map

A defense is easier to build when you start from the statute. Under the DTSA, an owner of a trade secret that relates to a product or service in interstate or foreign commerce may sue. 18 U.S.C. § 1836(b)(1). A "trade secret" is information of nearly any kind if two things are true: the owner "has taken reasonable measures to keep such information secret," and the information "derives independent economic value, actual or potential, from not being generally known to, and not being readily ascertainable through proper means by, another person who can obtain economic value from the disclosure or use of the information." 18 U.S.C. § 1839(3). "Misappropriation" means acquiring a secret knowing or having reason to know it was acquired by improper means, or disclosing or using it without consent in several described circumstances. 18 U.S.C. § 1839(5).

Nearly every state has adopted a version of the Uniform Trade Secrets Act (UTSA), and federal courts routinely analyze the two regimes together because the definitions track each other so closely. InteliClear, LLC v. ETC Global Holdings, Inc., 978 F.3d 653 (9th Cir. 2020). The DTSA does not displace state law, 18 U.S.C. § 1838, so most complaints plead both.

Each element creates a line of attack:

Element the plaintiff must prove The defense attack When it usually gets decided
A specifically identified secret Failure to identify; overbreadth Discovery sequencing, preliminary injunction, summary judgment
Not generally known Prior art, trade literature, general industry practice Summary judgment if the source is clear; otherwise trial
Not readily ascertainable (DTSA and most states) Time, cost, and availability of honest routes Often fact-bound; expert testimony
Reasonable measures NDA gaps, unmarked disclosures, open access Rarely on the pleadings; often a jury question
Independent economic value from secrecy No one would pay for the secrecy itself Summary judgment possible
Acquisition, disclosure, or use by improper means Reverse engineering, independent derivation, no knowledge, no use Summary judgment if similarity is weak; otherwise trial
Timeliness Three-year discovery limit Summary judgment, or JMOL as in Insulet

The bargain behind the weakness

Trade secret law is weak on purpose, and the reason is constitutional. Patent law gives a true monopoly for a limited time in exchange for public disclosure. States cannot hand out something similar for free. When Ohio's trade secret law was challenged as conflicting with the patent system, the Supreme Court upheld it in Kewanee because it left the public free to discover the same thing. A trade secret law, the Court wrote, "does not offer protection against discovery by fair and honest means, such as by independent invention, accidental disclosure, or by so-called reverse engineering, that is by starting with the known product and working backward to divine the process which aided in its development or manufacture." It also called trade secret protection "far weaker" in many respects than patent protection, and that weakness was central to its survival.

Fifteen years later, Bonito Boats, Inc. v. Thunder Craft Boats, Inc., 489 U.S. 141 (1989), struck down a Florida statute that banned copying boat hulls by the direct molding process. Reverse engineering a product in general circulation, the Court said, is one of the things only a patent can stop, and it "has never been a part of state protection under the law of unfair competition or trade secrets." The Court quoted a Ninth Circuit locksmith case on the point, which is worth reading for its own sake.

In Chicago Lock Co. v. Fanberg, 676 F.2d 400 (9th Cir. 1982), a father and son compiled a two-volume book of key codes for Chicago's "Ace" tubular locks by soliciting the data from locksmiths, who had derived it by picking customers' locks and grinding replacement keys. The trial court enjoined the book as an unfair use of the company's secret, reasoning that the locksmiths owed duties to their customers and the customers owed duties to the manufacturer. The Ninth Circuit reversed. A lock owner owes the manufacturer no duty about his own lock, so there was no breach anywhere in the chain. The court's sentence is the one to remember: "it is the employment of improper means to procure the trade secret, rather than mere copying or use, which is the basis of liability." Imposing a duty on purchasers, it added, would convert a trade secret into something like a patent, which federal law would not tolerate.

For a defendant, this history does practical work. It gives you a theme that judges respect, because it comes from the Supreme Court. The law wants competitors to study each other's products and race to the same answer. A plaintiff who is really saying "nobody else may make this" is asking for a patent it does not have.

Attack one: make the plaintiff say what the secret is

A plaintiff cannot prove that you took something it cannot describe. Courts require identification with enough particularity to separate the secret from general knowledge in the trade. Imax Corp. v. Cinema Technologies, Inc., 152 F.3d 1161 (9th Cir. 1998). MC Law's article on Identifying Trade Secrets in Litigation: Reasonable Particularity, Sequencing Discovery, and Protecting the Secret While You Sue covers the procedural detail, so only the defensive points follow.

First, identification drives every other defense. You cannot chart a secret against the prior art, or against your commit history, until you know what it is. That is why the defense usually asks for identification before producing its own technical documents, and why the plaintiff resists. California codifies the sequence in Code of Civil Procedure § 2019.210.

Second, do not overestimate what a federal court will do with a vague list early in the case. In Quintara Biosciences, Inc. v. Ruifeng Biztech, Inc., No. 23-16093 (9th Cir. Aug. 12, 2025), the Ninth Circuit reversed a district court that had struck most of a plaintiff's alleged secrets under Rule 12(f) for insufficient particularity at the outset of discovery. The DTSA, it said, contains no California-style requirement to identify with particularity from the start, and whether a secret is sufficiently particular is a question of fact that usually waits for summary judgment or trial. Courts can still manage the order of discovery under Rule 26 and Rule 16, but a motion to end the claim early on this ground now faces an obstacle in that circuit.

Third, the argument gets stronger as the case goes on. Cases are lost at summary judgment for this reason. In IDX Systems Corp. v. Epic Systems Corp., 285 F.3d 581 (7th Cir. 2002), a 43-page description of a software package failed because it did not separate secrets from everything else in the product, and many items were visible to any user. Judge Easterbrook wrote that "things that any user or passer-by sees at a glance are 'readily ascertainable by proper means.'" In NEXT Payment Solutions, Inc. v. CLEAResult Consulting, Inc., No. 24-1377 (7th Cir. Jan. 13, 2026), the plaintiff listed thirty-four software "modules" with names such as "Online Self Scheduling." The Seventh Circuit affirmed summary judgment, quoting the district court's remark that describing software functions without the underlying methods "is like saying someone stole your top secret apple pie recipe, but never identifying the secret recipe itself." It also made a point defendants should remember: that a tool sat behind a login did not make its functions secret, and any confidentiality agreement covering users went to misappropriation, not to whether a secret existed at all. The court was careful to say it was not holding that software plaintiffs must always produce source code.

Fourth, identification is often the fastest way to beat a preliminary injunction. In Insulet Corp. v. EOFlow, Co., 104 F.4th 873 (Fed. Cir. 2024), the Federal Circuit reversed an injunction that defined the secrets as any confidential information, called that definition "severely overbroad," and criticized a "hazy grouping of information." The Third Circuit vacated an injunction on similar grounds in Mallet & Co. v. Lacayo, 16 F.4th 364 (3d Cir. 2021), tying the point to Rule 65(d)'s demand that injunctions describe restrained conduct in reasonable detail.

Two related procedural points: compile a record of each shifting definition, because seven failed attempts helped produce a large fee award in Gabriel (discussed below); and take a Rule 30(b)(6) deposition early. In Stutz Motor Car of America, Inc. v. Reebok International, Ltd., 909 F. Supp. 1353 (C.D. Cal. 1995), the plaintiff's corporate designee testified that the secret was the air bladder and only the air bladder. That testimony, combined with a patent disclosing the air bladder, was the beginning of the end of the case.

Attack two: generally known and readily ascertainable

These two phrases are usually said together, but they do different work.

Generally known asks what the relevant people already know. "Information that is public knowledge or that is generally known in an industry cannot be a trade secret." Ruckelshaus v. Monsanto Co., 467 U.S. 986 (1984). The relevant audience is the trade, and the statute says so: the comparison is to another person "who can obtain economic value" from the information. A process unknown to the general public but familiar to every formulator in the field is not secret.

Readily ascertainable asks what those people could easily find out by honest means, even if they have not done so yet. The UTSA comment gives the classic examples: trade journals, reference books, and published materials. The Fourth Circuit quoted that language from South Carolina's version in Hoechst Diafoil Co. v. Nan Ya Plastics Corp., 174 F.3d 411 (4th Cir. 1999).

The California difference, and why it changes the proof

The UTSA's definition contains the words "not being readily ascertainable by proper means." The California Legislature deleted them when it adopted Civil Code § 3426.1(d). The Court of Appeal explained the consequence in ABBA Rubber, a customer-list case: "under California law, information can be a trade secret even though it is readily ascertainable, so long as it has not yet been ascertained by others in the industry." Readily ascertainable status is not part of the definition, but a legislative committee comment preserved it as a defense, and California's pattern instructions treat it as an affirmative defense (CACI No. 4420).

That sounds like a technicality. It changes who carries what, and it may change what you must prove.

Under the DTSA and in states that kept the uniform language, ascertainability is part of what makes something a trade secret, so the plaintiff has the burden. And if information really is readily ascertainable, there is no secret, whichever route the defendant took. The Third Circuit stated that rule under pre-UTSA Pennsylvania law in SI Handling Systems, Inc. v. Heisley, 753 F.2d 1244 (3d Cir. 1985), saying a product susceptible to reverse engineering was unprotected "regardless of whether appellants in fact went through such an exercise or, as would be more likely, relied on their memory."

In California, ABBA framed the defense more narrowly. The defendants could win at trial if they showed that certain manufacturers obviously used rubber rollers, that those manufacturers were easily identified, and that "the defendants' knowledge of the plaintiff's customers resulted from that identification process and not from the plaintiff's records." That last clause suggests that a California defendant who actually used the plaintiff's records cannot rely on the fact that it could have found the information elsewhere. Lawyers should read their own jurisdiction's statute before assuming the uniform rule. California is the most significant departure, and federal courts in California apply the DTSA's text to the DTSA count, so one case can involve both standards. A jury may need separate instructions.

How "readily" gets tested

"Could have been figured out" is not the same as "readily." The word carries the weight, and courts look at time, cost, and practicality.

SI Handling itself shows how this fails. The defendants offered an engineering expert who said a small team could reverse engineer the system in ten to twelve weeks. The trial court found his testimony unreliable, and the Third Circuit agreed, remarking that it detected "the smell of the lamp about it." Even if credited, the opinion paled beside evidence that nobody had marketed a comparable product and that the defendants had not hired inexperienced engineers to build it from scratch. An expert's estimate of what could have been done, set against what no competitor ever did, is weak evidence.

The Second Circuit made a similar point about an algorithm in Vermont Microsystems, Inc. v. Autodesk, Inc., 88 F.3d 142 (2d Cir. 1996). The defendant argued that a triangle shading routine was just a variation of well-known Gouraud shading. That did not save it: a developer had produced a nearly identical routine within weeks of joining Autodesk, and the possibility that others might independently create their own versions did not strip the plaintiff's version of its status.

Good "readily ascertainable" evidence is concrete. It shows the specific handbook, the specific supplier datasheet, or the actual time a neutral engineer took. A declaration that says "any competent chemist could do this" is an invitation to cross-examination.

Combinations: the reason a stack of public documents may not be enough

Here is where defendants most often overreach. Public availability of the components does not necessarily defeat the whole. "A trade secret can exist in a combination of characteristics and components, each of which, by itself, is in the public domain," the Seventh Circuit said in Pribyl, while conceding that 3M's manuals surely could not protect instructions on assembling a cardboard box. The Second Circuit protected a "winning combination" of generic utility programs in Integrated Cash Management Services, Inc. v. Digital Transactions, Inc., 920 F.2d 171 (2d Cir. 1990), and the Fifth Circuit held in Tewari De-Ox Systems, Inc. v. Mountain States/Rosen, L.L.C., 637 F.3d 604 (5th Cir. 2011) that information in a patent application was gone but unique combinations of those disclosed elements could still qualify. See also AvidAir Helicopter Supply, Inc. v. Rolls-Royce Corp., 663 F.3d 966 (8th Cir. 2011), which said that the public availability of some or even most of a compilation's contents is not dispositive.

The boundary with honest aggregation is easy to miss. In Compulife Software Inc. v. Newman, 959 F.3d 1288 (11th Cir. 2020), individual life insurance quotes were available to anyone on the website, but a hired bot pulled more than 43 million of them in four days. The magistrate judge reasoned that because each quote was public, there could be no misappropriation. The Eleventh Circuit vacated. "Public availability creates a vulnerability," the court wrote, but "taking enough of them must amount to misappropriation of the underlying secret at some point." It remanded to decide whether the scraping was improper means. Scraping law remains unsettled, and any defendant who collected public data at scale should take a hard look at how it was collected, including terms of use and whether the requests were disguised.

So the defensive question is sharper than "is it public?" It is: is the claimed combination itself public, or is there a real commercial reason that assembling it would be trivial? If the plaintiff identifies a combination, the defense needs a single reference, or a closely related set of references that people in the field actually use together, or evidence that this combination is a routine choice. Winning that fight usually requires an expert.

Attack three: public disclosure

Monsanto states the rule in property terms. If someone "discloses his trade secret to others who are under no obligation to protect the confidentiality of the information, or otherwise publicly discloses the secret, his property right is extinguished." The contested questions are what counts as public, how much was disclosed, and when.

Patents and published applications

The rule is firm. "After a patent has issued, the information contained within it is ordinarily regarded as public and not subject to protection as a trade secret." On-Line Technologies, Inc. v. Bodenseewerk Perkin-Elmer GmbH, 386 F.3d 1133 (Fed. Cir. 2004). Stutz calls it black letter law with roots in federal supremacy.

Publication now comes earlier than issuance. Under 35 U.S.C. § 122(b), applications are generally published promptly after 18 months from the earliest filing date for which a benefit is sought, unless the applicant certifies that it will not seek protection abroad in a system that publishes. That creates a fertile line of inquiry: file histories and foreign counterparts often disclose more than the plaintiff remembers. In Forcier v. Microsoft Corp., 123 F. Supp. 2d 520 (N.D. Cal. 2000), a Patent Cooperation Treaty application published fifteen or seventeen of the plaintiff's asserted secrets, and he had mailed copies to Microsoft, Apple, and IBM telling them the content was in "the public domain."

Three cases refine the rule.

Foreign patents count. In Ultimax, a California-law case, the claimed secret was the use of lithium carbonate and citric acid in a type of cement. A Japanese patent disclosed that combination. The plaintiff argued that how the defendant really got the information mattered. The Federal Circuit answered that "that is irrelevant if there is no secret," and held that "[i]nformation disclosed in a patent, even a foreign one, is 'generally known to the public,' especially the relevant public in the cement industry." The district court had also reasoned that publishing the general combination encompassed any narrower claim to a specific ratio, and the Federal Circuit affirmed. Plaintiffs will try to distinguish that by defining a specific ratio with real data behind it.

The wrongdoer's own patent counts going forward. In Attia, an architect alleged that Google had filed patent applications in 2011 containing his secrets, which were published in 2012. By the time the DTSA took effect on May 11, 2016, those applications were public. He sued under the DTSA anyway. The Ninth Circuit agreed that pre-enactment misappropriation does not by itself preclude a claim for continued use after enactment, but held that the publication "placed the information in the public domain and necessarily extinguished its trade secret status," and it rejected his equitable estoppel argument. That does not necessarily leave an owner without a remedy for the earlier wrong (Attia had sued on state law theories in 2014), but later use of what is now public is another matter. BondPro Corp. v. Siemens Power Generation, Inc., 463 F.3d 702 (7th Cir. 2006), confronted a defendant's published application. It explained that published applications are studied by inventors in the field, so a secret disclosed in them "will ordinarily (we need not decide whether invariably) lose its status."

The disclosure only reaches what it discloses. This is the plaintiff's best reply. In On-Line, the defendant's post-patent use was not misappropriation "unless the activities related to information not disclosed in the patent." Know-how that goes beyond the specification, such as process conditions, tolerances, or test results, may survive. In Tewari, the combinations survived. So patent defenses must be charted line by line. MC Law's Trade Secret Defense Toolkit: Independent Development Records, Public Domain Charts, and Bad Faith Fee Motions includes a chart for this.

Timing also matters in a way that sometimes produces an outright win. On-Line and Stutz both ended in judgment for the defendant on the trade secret claim because the plaintiff could not show use before the patent issued. In Stutz, the court found that Reebok did not use the claimed secret, if at all, until after the patent issued in September 1986. A chronology of your development against the plaintiff's first publication is one of the first things to build.

Posting, court files, and the "no one noticed" problem

Publication is a question of fact, and not every appearance in a public place ends secrecy. "The mere fact of mention in a public document is not controlling," the Seventh Circuit said in BondPro, because "no one may have noticed the document."

The leading contrast is between Hoechst and the Scientology cases. In Hoechst, a 28-page description of a film process had been inadvertently filed unsealed in another lawsuit, and a lawyer for the defendant found it in the courthouse. The Fourth Circuit held that presence in a court's public file, "in and of itself," did not make the information generally known or readily ascertainable. Trade journals are read by the trade. One clerk's office is not. In Religious Technology Center v. Lerma, 908 F. Supp. 1362 (E.D. Va. 1995), documents sat in an open court file for an extended period and were also posted to the Internet, and the court held them no longer secret. Religious Technology Center v. Netcom On-Line Communication Services, Inc., 923 F. Supp. 1231 (N.D. Cal. 1995), reasoned that Internet posting can make material generally known, at least to the people interested in the subject.

California's DVD Copy Control Ass'n v. Bunner, 116 Cal. App. 4th 241 (2004), put the question in technical terms. The DeCSS program had spread across the Internet before the defendant reposted it, and the court concluded that a preliminary injunction could not be justified on a record suggesting the information had already become widely known. Secrecy is relative, and a brief or obscure posting may not destroy it. Wide and persistent posting will.

The practical lesson for defendants is to find out who actually saw the disclosure. Download counts, forum replies, citations, and mirrors are evidence. So is a former colleague's testimony that everyone in the field read the same paper. And if the plaintiff itself published, say so loudly. In CVD, Inc. v. Raytheon Co., 769 F.2d 842 (1st Cir. 1985), Raytheon had described its process in periodic reports to the government, available to the public by 1979. A jury found that it had no secret, knew it, and had asserted secrecy in bad faith to extract a license from two former employees, which supported antitrust liability.

For protective-order issues on the other side of this problem, see Protective Orders, Confidentiality Designations, and Sealing: Secrecy and Its Limits in Federal Court. Trial use deserves attention too. In Littlejohn v. BIC Corp., 851 F.2d 673 (3d Cir. 1988), a party that did not object when confidential documents were used as trial exhibits was held to have waived protection under the protective order for the exhibits actually admitted. That cuts both ways, and it is one more reason to follow what your opponent does with its claimed secrets during a prior lawsuit.

Sales, demonstrations, and things a customer can see

A product sold on the open market discloses whatever a buyer can learn from it. That is the point of the reverse engineering rule, and IDX and NEXT Payment apply the same logic to software functions visible to users. Plaintiffs answer with Trandes Corp. v. Guy F. Atkinson Co., 996 F.2d 655 (4th Cir. 1993): source code was still "completely secret," and a jury could find object code secret too, where only two complete licenses had been granted under use restrictions and a small advertisement for a demonstration disk produced no sales. Distributing compiled software does not reveal the source, and a limited, licensed distribution may not even expose the binary.

So ask what the sale reveals. A shipped mobile application with unobfuscated business logic discloses more than a hosted service. A chemical product can be analyzed; whether analysis reveals the process used to make it is often the real dispute.

Attack four: reasonable measures

This is the argument defendants like most and win least often at an early stage.

Measures must be reasonable under the circumstances, and not heroic. Judge Posner explained why in Rockwell Graphic Systems, Inc. v. DEV Industries, Inc., 925 F.2d 174 (7th Cir. 1991). Rockwell gave thousands of engineering drawings to outside vendors under confidentiality agreements, and summary judgment had been granted against it. The Seventh Circuit reversed. Every dollar spent on security has a cost, and restricting access to drawings makes the work harder, "and therefore perfect security is not optimum security." The question was for a jury. The old Ninth Circuit case that speaks of relative secrecy is similar: in K-2 Ski Co. v. Head Ski Co., 506 F.2d 471 (9th Cir. 1974), displaying a ski cut lengthwise at a trade conference and allowing limited plant tours in a remote location did not defeat secrecy.

Where does this argument succeed? In three settings, mainly.

The owner's own paper says otherwise. Convolve, Inc. v. Compaq Computer Corp., 527 F. App'x 910 (Fed. Cir. 2013), is the case every engineer should hear about. The NDA required that oral and visual disclosures be identified as confidential at the time and confirmed in writing within twenty days. Convolve did not send confirmations for several disclosures, argued the parties' course of conduct excused them, and the court read the agreement's plain language against it. Marking clauses are not formalities. If the plaintiff signed one with you, get every disclosure and every follow-up letter.

No confidentiality at all. Ordinary NDA gaps, disclosures to customers without restrictions, and open file shares are classic evidence. The law has some give, since a confidential relationship can sometimes be implied. Direct Technologies, LLC v. Electronic Arts, Inc., 836 F.3d 1059 (9th Cir. 2016), declined to rule out the idea that a prototype given to a business partner without express restrictions might be reasonably protected. The Ninth Circuit affirmed judgment for the defendant on a different ground, finding no evidence of value in the secrecy of a design made for one short-lived project. That independent value argument is underused.

The Internet. Where the information was served to anonymous users with no controls, Compulife's "vulnerability" sentence gives you something to work with.

Expect the question to be treated as one of fact. Use the evidence anyway, because it colors everything else. A jury that hears a company's "crown jewel" sat on an unlocked share drive for a decade will view the rest of its story differently. For how the plaintiff's side thinks about these measures, see Building a Trade Secret Protection Program from Scratch.

The defenses the law names: reverse engineering and independent derivation

What the statute protects

Section 1839(6)(B) excludes reverse engineering, independent derivation, and "any other lawful means of acquisition" from improper means. The UTSA's comment lists the proper means: independent invention, reverse engineering, licensing, observation in public use or on display, and published literature. The comment adds that the known product must itself be acquired by fair and honest means, such as a purchase on the open market. The Restatement (Third) of Unfair Competition, §§ 39 through 45, treats the subject in depth, and Pamela Samuelson and Suzanne Scotchmer's "The Law and Economics of Reverse Engineering," 111 Yale L.J. 1575 (2002), explains why the rule makes economic sense in most industries.

The important limit is that the starting point has to be clean. The classic examples of improper means in Kewanee include theft, wiretapping, and "even aerial reconnaissance." A competitor's product purchased from a distributor is clean. A prototype obtained under an NDA for evaluation is not, and neither is a sample lifted from a shared lab.

Who has to prove what

Here plaintiffs and defendants often speak past each other. Direct evidence of copying is rare, so courts let plaintiffs prove use circumstantially, by showing access to the secret and similarity between it and the defendant's product. That inference can reach a jury. The defendant then has to explain.

But the burden of proof does not move. Sargent Fletcher is the case to read. An aerospace manufacturer asked for an instruction that once the jury found disclosure and a substantially similar product, "the burden then shifts" to the defendant to prove it developed independently. The Court of Appeal affirmed the refusal. "The law deals with this problem by shifting the burden of producing evidence, not the burden of proof." Able had pleaded 14 affirmative defenses, none of them reverse engineering or independent derivation. Both sides agreed those are not affirmative defenses, and the court treated them as a denial of the plaintiff's own element, not as new matter. (The trial lasted seven weeks and the jury deliberated for five. These cases can be long.) A careful defendant will nonetheless plead independent development to avoid any waiver argument, while maintaining that the plaintiff retains the burden.

The inference also has limits. In Stratienko v. Cordis Corp., 429 F.3d 592 (6th Cir. 2005), an inventor sent a catheter proposal to Cordis, whose policy channeled outside submissions away from its engineers "to minimize exposing our innovative engineering talent pool to external ideas." Cordis later sold a similar catheter. The plaintiff lost because his evidence failed "to identify which, if any, innovative features" the two designs shared. Similarity in features that are common or dictated by function proves nothing. The court also accepted declarations from Cordis employees despite their self-interest, which matters because in these cases the only people who know the facts usually work for the defendant.

Plaintiffs have a strong rejoinder on similarity: differences do not automatically help. "[T]he user of another's trade secret is liable even if he uses it with modifications or improvements upon it effected by his own efforts, so long as the substance of the process used by the actor is derived from the other's secret." Mangren Research & Development Corp. v. National Chemical Co., 87 F.3d 937 (7th Cir. 1996). Vermont Microsystems likewise holds that a plaintiff need only show substantial derivation, not identity. And "use" is read broadly. Oakwood Laboratories LLC v. Thanoo, 999 F.3d 892 (3d Cir. 2021), adopted an expansive reading of the term at the pleading stage, in a case about a competitor's development program.

Speed, errors, and other tells

Because the evidence is circumstantial, certain facts carry disproportionate weight.

Speed. How fast did you get there? In Integrated Cash, the new company had a prototype within two weeks of the programmers' arrival. In Vermont Microsystems, the routine appeared within weeks. If your development really took three years and failed repeatedly, that is your single most persuasive fact.

Shared mistakes. Nothing proves copying like a copied error. In Motorola Solutions, Inc. v. Hytera Communications Corp., 108 F.4th 458 (7th Cir. 2024), the Seventh Circuit recounted proof that "minor coding errors in Motorola's code appeared in exactly the same spots in Hytera's code," and that Hytera had altered code to hide its origin. Independent developers make their own mistakes. Forensic code comparisons look for this kind of thing: identical typos in comments, the same off-by-one bug, identical magic constants, and identical dead code. Formulation chemists look for the same pattern in a trace ingredient that serves no function.

Volume of downloads. The cases with terrible facts for defendants have logs. Motorola involved thousands of files, and in Epic Systems Corp. v. Tata Consultancy Services Ltd., 980 F.3d 1117 (7th Cir. 2020), a services firm downloaded thousands of documents from a customer portal over two years. If you cannot explain the logs, change your theory or settle early. See Proving or Defending Against Trade Secret Damages: A Practical Guide for what happens to the money next.

Could have is not did

A common mistake is to tell a court that you could have reverse engineered the product when you did not. Under SI Handling and the definitional approach, truly easy reverse engineering means no secret existed. But if reverse engineering was merely possible with effort, the fact helps very little, and your witnesses will have to admit they did not do it. Courts treat a real project differently from a retrospective estimate. If your team really bought the competitor's product and tore it down, your best evidence is the purchase receipt, the teardown notes, the instrument reports, and the time sheets.

Contract can change the answer

The statute protects reverse engineering as a lawful means. A contract can make it unlawful between the parties. Chicago Lock rejected an implied duty on purchasers, but it did not involve an express promise. Software license terms commonly prohibit reverse engineering, and "breach ... of a duty to maintain secrecy" is itself improper means under § 1839(6)(A). In Bunner, the plaintiff's theory was exactly that: the DeCSS code had been obtained by reverse engineering software whose click-wrap license prohibited it.

So before a teardown, read the terms. Separately, copyright law allows some reverse engineering of software for interoperability. Sega Enterprises Ltd. v. Accolade, Inc., 977 F.2d 1510 (9th Cir. 1992). That does not automatically resolve a contract or trade secret claim. MC Law's Copyright in Software Interfaces After Google v. Oracle: APIs, Compatibility, and Reverse Engineering and Software Interoperability and Reverse Engineering Clearance Checklist: A Practical Checklist cover those issues.

The engineering record: what actually happens in a repository, a lab, and a clean room

For many defendants the outcome will turn on whether development records can be trusted. So it helps to know what those records are at the system level.

What git proves and what it does not

A git repository is a chain of content-addressed objects. Each commit stores a snapshot of the tree, the hash of its parent, an author and committer name, and two timestamps, and its own identity is a hash of all of that. Change anything in an old commit and every later hash changes. That gives you a tamper-evident structure: once a third party has recorded a commit hash, nobody can quietly rewrite what came before it.

The weakness is that the timestamps and names are supplied by the machine that created the commit. Anyone can set them, and a git rebase or git filter-repo can rewrite an entire history on a laptop in seconds, producing a consistent chain with invented dates. Squash merges and force pushes destroy intermediate history in ordinary, innocent use. So opposing experts will ask a simple question: what evidence outside the repository anchors this history in time?

Good answers include:

  • Server-side records from the hosting platform, such as push events, pull request timestamps, and audit logs, which the developer cannot rewrite.
  • Continuous integration logs and artifact stores recording that a particular commit hash was built on a particular day.
  • Issue tracker entries, design documents, chat messages, and code review comments that mention the same work.
  • Backups, snapshots, and old release packages, which fix the state of the code at a date.
  • Package registry publications, and customer deliveries, whose hashes can be compared to the repository.
  • Signed commits and trusted timestamps, where they exist, although their absence means little because most teams do not use them.

What does independent development look like in that record? It looks messy. Abandoned branches, reverted commits, bugs fixed by trial and error, benchmarks that got better slowly, and design arguments in pull requests are all persuasive, because a person copying a finished design has no reason to generate them. A repository whose first commit contains 40,000 lines of working code is the opposite, though it may have an innocent explanation, such as a migration from another system. If so, go find the earlier system.

Preserve all of it as it exists now. Do not clean up. Do not squash. Clone with full history, capture the hosting provider's logs before retention periods expire, and have someone other than the accused engineer do the collection. The rule on loss of electronically stored information, Rule 37(e), applies once litigation is reasonably anticipated, and a demand letter nearly always triggers that duty. Zubulake v. UBS Warburg LLC, 220 F.R.D. 212 (S.D.N.Y. 2003). See Litigation Holds, Spoliation, and Rule 37(e): Preserving Electronic Evidence Before It Costs You the Case.

Getting the records admitted is usually manageable. Fed. R. Evid. 901 allows authentication by evidence describing a process or system and showing that it produces an accurate result. Rule 902(13) and (14) allow certification of electronic records and of data copies verified by hash values. Rule 803(6) covers records of regularly conducted activity, though counsel should expect a fight over whether a developer's commit message counts. For the mechanics, see Hearsay and Business Records at Trial: Rule 803(6), Self-Authentication, and Getting Documents Into Evidence.

Formulas: what the same proof looks like in a lab

Chemistry has its own version of a commit history. Dated laboratory notebooks with witness signatures, batch records, retained samples, supplier purchase orders, certificates of analysis, and test reports do the same work. Purchase orders are especially strong because a vendor holds the matching record: you cannot have been testing an additive in 2023 if you first bought it in 2025. Failed batches matter as much as successful ones.

A reverse engineering project leaves a distinctive trail, too: a retail receipt for the competitor's product, chain of custody, an outside laboratory's gas chromatography and mass spectrometry reports, and the iterations that followed. Be careful about what analysis actually shows. Compositional analysis may reveal ingredients without revealing the order of addition, the cure schedule, or the purity grade that makes the difference, and those details may be the alleged secret.

Clean rooms

A clean room is an organizational structure for producing compatible or competing work without exposure to someone else's protected material. The classic model separates a "dirty" team, which studies the target and writes a functional specification, from a "clean" team, which has never seen the target and builds only from that specification. A lawyer or technical reviewer screens the specification so that it describes required behavior and not protected expression or secret implementation details. Phoenix Technologies' reimplementation of the IBM PC BIOS in the 1980s is the famous example.

The best-reported judicial example is Computer Associates International, Inc. v. Altai, Inc., 982 F.2d 693 (2d Cir. 1992). Altai learned that one of its programmers had brought in code copied from his former employer. On advice of counsel, it excluded him from the project, locked away his copy of the code, and gave eight other programmers, none of whom had worked on the earlier version, a description of the operating system services they needed. The rewrite took about six months, and the new version was held not to infringe the copyright. The Second Circuit nonetheless remanded the trade secret claim for a merits determination, including a theory based on the original acquisition. A rewrite limits what happens next. It does not erase what already happened.

Clean room records need to prove the negative: who was walled off, what each person had seen before, what went into the specification, and how questions were resolved. Teams should keep a log of every question from the clean side and every answer from the dirty side, because a stray message can undo the structure. Generative coding tools complicate this if the tool had access to contaminated repositories or prompts. See Trade Secrets and Generative AI: Reasonable Measures, Prompt Leakage, and Vendor Terms.

No improper acquisition, no knowledge, no use

Some defenses do not turn on whether the plaintiff has a secret at all.

The pie. Silvaco Data Systems v. Intel Corp., 184 Cal. App. 4th 210 (2010), disapproved on another ground by Kwikset Corp. v. Superior Court, 51 Cal. 4th 310 (2011), held that a company running compiled software allegedly built with stolen source code did not thereby use the source code. "One who bakes a pie from a recipe certainly engages in the 'use' of the latter; but one who eats the pie does not, by virtue of that act alone, make 'use' of the recipe in any ordinary sense." A customer that never saw the secret lacks the required knowledge. This matters to companies that buy tools from a vendor accused of copying.

No duty. Chicago Lock shows how to walk the chain of duties link by link. For third parties, the statute requires knowledge or reason to know. 18 U.S.C. § 1839(5). Receiving a demand letter can change that going forward, a point the Hiring From a Competitor: Inevitable Disclosure, Employee Raiding, and the New Employer's Trade Secret Exposure article takes up. Once you have notice, continued use needs a considered decision.

Possession and planning are not misappropriation. A former employee with old files is a concern, and the files may need to be returned, but having knowledge is different from using it. In FLIR Systems, Inc. v. Parrish, 174 Cal. App. 4th 1270 (2009), the trial court found neither misappropriation nor threatened misappropriation where departing engineers had a business plan but no funding, no employees, no customers, and no product.

General skill and knowledge

The law has long separated what belongs to the employer from what belongs to the person. In AMP Inc. v. Fleischhacker, 823 F.2d 1199 (7th Cir. 1987), an employee "is free to take with him general skills and knowledge acquired during his tenure with his former employer," and six pages listing hundreds of pieces of internal information did not identify anything else. CVD treats the principle as serving labor mobility and competition. SI Handling, adopting a Pennsylvania phrase, declined to "act as a judicial eraser to blot out ... knowledge and skill."

The line is real but not generous. Memorizing does not launder a secret. "The form of information, whether written or memorized, is immaterial under the trade secrets statute." Ed Nowogroski Insurance, Inc. v. Rucker, 971 P.2d 936 (Wash. 1999). An expert who explains why a technique is textbook will do far more good than a witness who says she just happened to remember the numbers. Specific constants, recipes, customer data, and process parameters tend to sit on the employer's side. Approaches, instincts about what will not work, and familiarity with the field tend to sit on the employee's side, although a former employee's knowledge of what failed can be contested as negative know-how.

Federal law reinforces the point at the remedy stage. An injunction under the DTSA may not prevent a person from entering an employment relationship, and any conditions must rest on evidence of threatened misappropriation and "not merely on the information the person knows." 18 U.S.C. § 1836(b)(3)(A)(i)(I). California rejected inevitable disclosure outright in Whyte v. Schlage Lock Co., 101 Cal. App. 4th 1443 (2002). Other jurisdictions follow PepsiCo, Inc. v. Redmond, 54 F.3d 1262 (7th Cir. 1995), to differing degrees. That doctrine and the new-employer side are covered in Hiring From a Competitor: Inevitable Disclosure, Employee Raiding, and the New Employer's Trade Secret Exposure and Lateral Hire Toolkit: Candidate Certifications, Onboarding Instructions, and Cease-and-Desist Responses.

Contracts as the real battlefield

Many claims are really about a contract, and the contract may help you. NDAs commonly exclude information already known to the recipient, independently developed, or publicly available. On-Line's NDA excluded information that "becomes publicly available other than by breach." Convolve turned on marking terms. Some NDAs contain residuals clauses, and many expire on a stated date.

Contract can also hurt. A contract claim may survive when the tort claim fails, because it does not require a secret. UTSA displacement provisions typically preserve contract remedies. Angelica Textile Services, Inc. v. Park, 220 Cal. App. 4th 495 (2013). And federal patent policy does not stop parties from agreeing to pay for unpatented ideas. Aronson v. Quick Point Pencil Co., 440 U.S. 257 (1979). An accused party with a broad covenant has to analyze it separately. See Drafting a Nondisclosure Agreement Checklist: A Practical Checklist and Non-Compete Agreements Under Siege: FTC Rulemaking and State Law Developments.

Time: the limitations defense after Insulet

The DTSA gives three years from the date the misappropriation "is discovered or by the exercise of reasonable diligence should have been discovered," and "a continuing misappropriation constitutes a single claim." 18 U.S.C. § 1836(d). California's Civil Code § 3426.6 and most UTSA enactments use similar language.

A trade secret plaintiff normally pleads suspicion and argues that it could not have known until discovery. Defendants should go looking for what the plaintiff knew and when, because this is often the cleanest route to judgment.

Insulet is now the most important reading. Insulet made a wearable insulin pump. Its former employee consulted for EOFlow in 2018. At a June 2018 trade show, an Insulet employee emailed management: "EOFlow has cloned our product . . . Our legal team is aware of this. Someone should request samples." Another wrote that "[w]e need to see if this solution is based on our IP." Insulet's complaint came in August 2023. A jury found no time bar and awarded $170,000,015 in compensatory damages and $282,000,005 in exemplary damages, later reduced by the trial court to $25.8 million and $33.6 million.

The Federal Circuit reversed outright. The majority did not resolve whether the DTSA uses inquiry notice, under which the clock starts when a reasonable person would begin investigating, or the stricter standard of Merck & Co. v. Reynolds, 559 U.S. 633 (2010), under which it starts when a diligent plaintiff would have discovered the facts. It held the claim late even under Merck. Several points deserve attention:

  • The test is what was needed to plead. The court asked whether Insulet knew or should have known facts sufficient to allege access and similarity, and it looked at Insulet's own eventual complaint to see what kind of knowledge it had thought was enough.
  • Access is satisfied by knowing where the employee went. Following Seatrax, Inc. v. Sonbeck International, Inc., 200 F.3d 358 (5th Cir. 2000), the court treated knowledge that a former employee was working for the competitor on the product as notice of access.
  • One relationship, one clock. The jury had been told to consider limitations separately for each secret. The court held that the secrets, all passed by one person in one course of conduct, formed a single claim. Under its reasoning, if the period has run on a secret, "the statute of limitations has also run with respect to related, broader trade secrets of which it is a part." The Colorado Supreme Court's decision in Gognat v. Ellsworth and the district court's in Intermedics v. Ventritex supplied much of this reasoning.
  • It is contested. Judge Prost dissented, arguing that the majority's application of the discovery rule blurs into inquiry notice and encourages plaintiffs "to race to the courthouse with undeveloped claims based on mere suspicion." The Federal Circuit was applying a general federal statute without First Circuit authority on point, so other circuits are free to differ. Insulet petitioned for rehearing en banc in July 2026, so check the docket before relying on the panel decision.

Other authorities support an early start. CMI Roadbuilding, Inc. v. Iowa Parts, Inc., 920 F.3d 560 (8th Cir. 2019), applying both statutes, relied on Iowa inquiry-notice language that "[t]he plaintiff need only know a problem exists." Cadence Design Systems, Inc. v. Avant! Corp., 29 Cal. 4th 215 (2002), holds that continued use against the same defendant is part of one claim accruing at the initial misappropriation. Cypress Semiconductor Corp. v. Superior Court, 163 Cal. App. 4th 575 (2008), starts the clock against a third party when the plaintiff has reason to suspect that party knows or should know the information is a secret. Stutz found constructive knowledge from wide advertising and media reports about Reebok's shoes. In Forcier, a single Wall Street Journal article prompted suspicion and a call to a lawyer, and six years passed before suit.

What should a defendant do with this? Demand the plaintiff's internal competitive intelligence, trade show reports, and first legal hold in discovery. Check when your product was publicly launched, photographed, or described in a prospectus. Sort the plaintiff's privilege log by date: its first entry mentioning you may be a limitations exhibit. Also calculate the critical date immediately, and preserve the defense by pleading it under Rule 8(c). Note too that the burden of proving the plaintiff's knowledge may be debated. In Insulet, the jury instruction put the discovery burden on the plaintiff once the defendant showed the first misappropriation came outside the period, but nobody challenged the instruction on appeal.

Preemption and displacement

A plaintiff often adds fraud, conversion, unfair competition, and unjust enrichment counts to a trade secret claim, in part because those counts carry different limitations periods and do not require a real secret. The UTSA was written to stop that. Section 7 of the uniform act displaces conflicting tort, restitutionary, and other state law providing a civil remedy for misappropriation, while preserving contract claims and remedies not based on misappropriation.

Courts disagree about how far this goes. California's approach is broad: Silvaco follows K.C. Multimedia, Inc. v. Bank of America Technology & Operations, Inc., 171 Cal. App. 4th 939 (2009), in holding that the act preempts claims based on the same nucleus of facts as misappropriation, reasoning that the statute's purpose would be subverted by leaving alternative bases for liability intact. Angelica Textile marks the limit: claims not based on misappropriation survive, including an employee's breach of duty on independent facts. In On-Line, the Federal Circuit affirmed displacement under Connecticut's act of a fraud claim that was at bottom about the trade secret. Wisconsin goes the other way on confidential information falling short of a secret. Burbank Grease Services, LLC v. Sokolowski, 717 N.W.2d 781 (Wis. 2006), holds that its statute does not preclude other civil remedies for information that does not meet the definition. Check your state before assuming either approach.

Note what does not preempt. The DTSA itself displaces nothing, so state claims rise or fall on state law. 18 U.S.C. § 1838.

Federal copyright preemption is a separate question. Under 17 U.S.C. § 301, a state claim survives if it has an extra element that makes it different in kind. Altai and Trandes both hold that a trade secret claim generally survives, because breach of a duty of confidence or improper acquisition is the extra element. Bare allegations of copying are more vulnerable, and non-trade-secret state claims may fall, as they did in Spear Marketing, Inc. v. BancorpSouth Bank, 791 F.3d 586 (5th Cir. 2015). For the full analysis, see Copyright Preemption Under Section 301: When Federal Law Wipes Out Your State Law Claim.

Two narrower shields: immunity and unclean hands

Whistleblower immunity

Section 1833(b) gives individuals immunity under any federal or state trade secret law for disclosing a trade secret in confidence to a government official or an attorney "solely for the purpose of reporting or investigating a suspected violation of law," or in a document filed under seal. 18 U.S.C. § 1833(b)(1). A retaliation plaintiff may also use the information in the proceeding with seal protections. An employer that does not give notice of this immunity in contracts governing confidential information, entered into or updated after the statute's enactment, cannot recover exemplary damages or fees from that employee. 18 U.S.C. § 1833(b)(3).

The courts have been careful with this defense. In Unum Group v. Loftus, 220 F. Supp. 3d 143 (D. Mass. 2016), the court refused to apply it at the pleading stage because the record did not show what was taken or what was done with it. In FirstEnergy Corp. v. Pircio, No. 1:20-cv-01966 (N.D. Ohio Mar. 8, 2021), the pleadings themselves showed disclosure only to counsel and, through counsel, the government, so the court could apply the immunity at the pleading stage, and it rejected the argument that an improper-acquisition claim based on exceeding authorized access survived anyway. And in Rozdilsky v. Liquidity Services, Inc., No. 22-3355 (D. Md. Feb. 20, 2026), summary judgment for the employee rested on unrebutted evidence of the limited disclosure and sole purpose, though a separate contract claim based on the same photographs had to be analyzed on its own. The statute does not authorize unlawful access to material. 18 U.S.C. § 1833(b)(5). See Whistleblower and Retaliation Claims: SOX, Dodd-Frank, the False Claims Act, and State Law.

Unclean hands

This one is narrower than clients hope. California allowed the defense against a technology conversion claim in Unilogic, Inc. v. Burroughs Corp., 10 Cal. App. 4th 612 (1992), where there was evidence that the plaintiff's predecessor had obtained the very contract by paying a $20,000 bribe to the defendant's employee. The court stressed that the misconduct "must relate directly to the transaction concerning which the complaint is made." That the plaintiff is a bully, or has itself hired from competitors, usually will not qualify. A plaintiff that obtained the claimed secret from your company, or that lied to you to get the relationship, might. The defense is equitable, so check how your jurisdiction treats it against damages claims, and check the pleading requirements. Rule 9(b) may apply if the theory is fraud.

Turning the tables: fees, sanctions, and related remedies

The statute allows a court to award reasonable attorney's fees to the prevailing party "if a claim of the misappropriation is made in bad faith, which may be established by circumstantial evidence." 18 U.S.C. § 1836(b)(3)(D). UTSA enactments are similar, including California Civil Code § 3426.4.

What "bad faith" means is not fully settled

The DTSA does not define it. Most courts borrow California's two-part test from Gemini Aluminum Corp. v. California Custom Shapes, Inc., 95 Cal. App. 4th 1249 (2002): objective speciousness and subjective bad faith. FLIR explains that objective speciousness "exists where the action superficially appears to have merit but there is a complete lack of evidence to support the claim." The test is less demanding than frivolousness, because a specious claim can look good on the surface. Direct Technologies applies it, and an unpublished Ninth Circuit decision assumes without deciding that it governs the DTSA. Gorovenko v. Activate Clean Energy LLC, No. 24-5840 (9th Cir. Oct. 14, 2025).

Other courts phrase things slightly differently. Recoop LLC v. Outliers Inc., No. 1:24-cv-01810 (S.D.N.Y. June 20, 2025), surveys the formulations in the Third, Fourth, Sixth, and Seventh Circuits, ranging from a complete lack of evidence plus knowing or reckless pursuit, to a claim with no chance of success, to a meritless claim maintained knowingly for an improper purpose. It then adopts the speciousness-plus-subjective-bad-faith test and denies fees. Read your circuit's formulation before you draft.

What wins

The winning fact pattern is usually persistence after the defect was exposed. "Made" includes maintained.

What loses

Losing on the merits is not bad faith, and a merely aggressive demand is not either.

Procedure and traps

The biggest trap is voluntary dismissal. A plaintiff that sees the end coming may seek dismissal without prejudice under Rule 41. In Dunster Live, L.L.C. v. LoneStar Logos Management Co., 908 F.3d 948 (5th Cir. 2018), the Fifth Circuit held that such a dismissal does not make a defendant a prevailing party. So a defendant should ask the court to condition dismissal on prejudice or fees. CRST Van Expedited, Inc. v. EEOC, 578 U.S. 419 (2016), a Title VII case, does establish that a defendant can prevail without a merits ruling, so long as the case is resolved in its favor.

Build the record ahead of the motion. Send a letter identifying the fatal defect, with the evidence, as soon as you can responsibly do so. Ask for the identification again. Consider a Rule 11 motion, which must be served 21 days before filing. Fee motions are generally due within 14 days after judgment under Rule 54(d)(2), unless a statute or order provides otherwise. Rule 11 and § 1836 address different things, and Rule 11 and Inherent-Power Sanctions: The Safe Harbor, Fee Awards, and Attorney Conduct covers the former.

Some defendants have other tools. A company facing a real threat may be able to seek a declaration of non-misappropriation under 28 U.S.C. § 2201 and MedImmune, Inc. v. Genentech, Inc., 549 U.S. 118 (2007), a route AvidAir tried before losing on the merits. See Declaratory Judgment Actions: Jurisdiction, Discretion, and the Race to the Courthouse. CVD shows that bad-faith assertion of a known non-secret can support antitrust liability, although the standard is high and the court was careful to protect genuine enforcement. And the DTSA gives a damages claim for wrongful or excessive ex parte seizure. 18 U.S.C. § 1836(b)(2)(G).

Preserving what you win

Trial creates its own forfeitures. In Harbor Business Compliance Corp. v. Firstbase.io, Inc., No. 25-1278 (3d Cir. Aug. 27, 2025), the Third Circuit concluded that a defendant had forfeited its challenge to the sufficiency of evidence on protectability, because its Rule 50(a) motion addressed only misappropriation. The court then assumed the secrets were protectable and affirmed liability. A defendant who wins everything except the jury's sympathy needs every element in the motion made before the case is submitted.

Injunctions deserve separate attention at the opposite end of the case. Irreparable harm is not presumed under a statute that makes injunctions discretionary. First Western Capital Management Co. v. Malamed, 874 F.3d 1136 (10th Cir. 2017); see eBay Inc. v. MercExchange, L.L.C., 547 U.S. 388 (2006). The Second Circuit reasons that a defendant who wants to use a secret profitably, without spreading it, has every incentive to keep it confidential, so damages will often suffice. Faiveley Transport Malmo AB v. Wabtec Corp., 559 F.3d 110 (2d Cir. 2009). Faiveley leaves room for a rebuttable presumption where a misappropriator threatens to spread the secret to a wider audience, so a defendant should show concretely that it has no intention of doing so.

Two worked hypotheticals

The sealant formula

Back to Tessaly. The Corvane letter claims "the Formula," and nothing more. What does good defense work produce in the first ninety days?

The chemist's old laptop was returned, but a personal cloud account holds a 2019 Corvane spreadsheet she forgot about. Counsel has a forensic vendor image it, walls it off, and confirms by search and metadata that it was not opened after her departure. That is a bad fact handled well. Purchase orders show that she first bought the key plasticizer eighteen months after leaving, from a catalog supplier. Notebooks record forty-one batches, thirty-eight of them failures. A 2016 conference paper by academic authors reports the same polymer and plasticizer family, and a Corvane patent application, published in 2021, discloses a ratio range overlapping hers.

The response letter does not argue the merits in detail. It denies misappropriation, asks Corvane to identify the claimed secret under an appropriate confidentiality arrangement, confirms preservation, and proposes that outside counsel and experts exchange information. When Corvane finally identifies a ratio range plus a post-cure schedule, the picture splits. The ratio range is in Corvane's own application. Under Attia and BondPro, that part is gone for the future, and under Tewari, the post-cure schedule may survive if it is not disclosed. On that remaining point, Tessaly has the failed batches showing how it arrived at a different schedule by experiment. And Corvane's sales team reported on Tessaly's product at a trade show thirty-eight months before the letter, which starts a limitations inquiry under Insulet.

Tessaly still has a risk. The spreadsheet gives Corvane an access story, and the chemist's eleven years give it another. But Tessaly now has a defense it can show to a judge, and Corvane has a fee exposure if it pushes on after seeing the record.

The routing engine

Suppose a logistics start-up called Marrowgate hires two engineers from Dunmere Systems. Fourteen months later, Dunmere sues under the DTSA, alleging its route optimization engine was taken. Marrowgate's repository looks clean, with its first commit dated nine months before the hires, but its core solver module first appears in a single commit of 18,000 lines made three weeks after they arrived.

What looks like a catastrophe may be innocent, since the team had developed the solver in a separate experimental repository. The first job is to find it. If the experimental repository exists, with server-side push logs and a year of messy benchmarks, the 18,000-line commit becomes a migration. If it does not, the company has a serious problem, and a forensic comparison of the two codebases becomes urgent. A privileged review by an outside expert can look for shared errors of the Motorola kind and tell the company whether it has an independent development case or needs an Altai-style rewrite plus an early settlement.

The two possible findings lead to two very different cases. The company should learn which one it has before it makes any factual statement to the other side. That is the reason to keep the investigation privileged under Upjohn Co. v. United States, 449 U.S. 383 (1981), and to retain the technical consultant through counsel, as in United States v. Kovel, 296 F.2d 918 (2d Cir. 1961). The step by step version of this work is in Building a Trade Secret Defense From the First Demand Letter: A Practical Guide.

What this means on Monday

If you have received a letter, start with preservation and counsel, in that order, and use the guide and checklist linked below for the rest.

If you are a company that has not received a letter, the same logic gives you a cheap insurance policy: keep your repositories unsquashed where you can, retain continuous integration logs and lab notebooks longer than you think you need them, document where new hires' knowledge came from, and write down why you made technical decisions. Those records are ordinary engineering hygiene until the day they become your defense. The Trade Secret Defense Checklist: A Practical Checklist turns this article into a desk list.

For deeper reading, the Federal Judicial Center's Trade Secret Case Management Judicial Guide (2023), edited by Peter Menell and others, shows how judges are told to manage these cases. The Sedona Conference's Commentary on the Proper Identification of Asserted Trade Secrets in Misappropriation Cases addresses identification, and Mark Lemley's "The Surprising Virtues of Treating Trade Secrets as IP Rights," 61 Stan. L. Rev. 311 (2008), makes the case for treating the doctrine as a limited property right. Case research on any of these points can start on UpLaw (uplaw.us).

This article describes the law generally and is not advice about your dispute, where the facts, the state statute, and the contract language can change the result. If a demand has arrived, a short conversation with a litigator who understands both the doctrine and the engineering is usually worth having before anyone replies.

Related Documents

Related Articles

Related Guides

Related Checklists

Related Toolkits