Understanding AI Distillation In The Trade Secret Context

May 12, 2026
AI, News, Trade Secret

For most of the past year, when AI companies have talked about distillation, they have talked about it as a Chinese problem. In February 2026, OpenAI told the House Select Committee on China that DeepSeek employees had developed methods to circumvent OpenAI’s access restrictions and programmatically harvest outputs to train a competing model. Anthropic’s public technical disclosure went further, naming DeepSeek, Moonshot AI, and MiniMax, and quantifying the activity: roughly 24,000 fraudulent accounts, more than 16 million exchanges, focused on extracting Claude’s reasoning, coding, and agentic capabilities.

By April, the White House Office of Science and Technology Policy had adopted a national-security framing, accusing China of running “deliberate, industrial-scale campaigns” to distill American frontier models. Throughout this timeline, the narrative was that of foreign actors using proxy infrastructure to extract capabilities the United States had spent billions to build.

Recently, however, that narrative got more complicated. On the stand in Musk v. Altman in the Northern District of California, Elon Musk admitted under cross-examination that his AI company, xAI, had “partly” distilled OpenAI’s models. “It is standard practice,” Musk said, “to use other AIs to validate your AI.” There were, by the reporters’ account, audible gasps in the courtroom.

Whether or not it is in fact standard practice, distillation is now a topic that trade secret practitioners should understand. The legal frameworks for how courts will treat it are not yet settled, and the questions it raises map onto familiar trade secret doctrine in ways that are sometimes intuitive and sometimes not.

What Is Distillation?

In the technical literature, knowledge distillation is a method developed in the mid-2010s for transferring capabilities from a large, expensive model (the “teacher”) into a smaller, cheaper model (the “student”). The student is trained to reproduce the teacher’s outputs across a broad range of inputs, with the result being a model that approximates the teacher’s behavior at a fraction of the size and cost. As Anthropic’s own disclosure acknowledges, distillation is “widely used and legitimate,” and frontier labs routinely distill their own models to create smaller, cheaper versions for production deployment.

The contested form of distillation, sometimes called adversarial or unauthorized distillation, takes the same technique and applies it to someone else’s model. The distiller does not need to steal weights or breach servers. Access to the teacher model’s API is sufficient. The distiller submits a large volume of carefully constructed queries, captures the outputs, and uses those outputs as training data for a competing student model. The technique converts what would otherwise be a public-facing inference service into a training corpus for a rival system.

The finances of this technique are compelling, to say the least. To take one example, ChatGPT-5 reportedly cost more than $2 billion to develop. DeepSeek’s R1, which OpenAI alleges was built in part through distillation of ChatGPT outputs, reportedly cost approximately $6 million of marginal training compute. 

The Trade Secret Theory

The question for trade secret lawyers is whether what gets extracted through distillation can be a trade secret, and whether the extraction can constitute misappropriation.

First, it is always worth starting from first principles. A trade secret must be information that derives independent economic value from not being generally known or readily ascertainable, and that the owner has taken reasonable measures to keep secret. 18 U.S.C. § 1839(3). Frontier model weights and the proprietary training methodologies that produced them comfortably satisfy these criteria, just as compiled software and machine-learning architectures have for years. See, e.g., Camilla A. Hrdy, Trade Secrecy Meets Generative AI, 100 Chi.-Kent L. Rev. 317 (2025). The harder question is whether the outputs of a model, accessed through a public API, can carry trade secret status. And if they can, whether systematically harvesting them to reconstruct the underlying capabilities is misappropriation, or merely a novel form of reverse engineering.

Of course, that distinction matters. Reverse engineering of a publicly available product is a textbook proper means of acquisition under both the DTSA and the UTSA. See 18 U.S.C. § 1839(6)(B). If a frontier model is made available to the public through an API, and a competitor pays the access fees and queries it to learn how it behaves, the surface analogy to traditional reverse engineering is obvious: buy the product, study how it works, and build a competitor.

On the other hand, traditional reverse engineering targets a finished good, the embodied capability. Distillation targets the model’s underlying reasoning processes, learned representations, and capability distributions. As one recent analysis framed it, the trade secret on this theory is “the aggregate of learned representations that required billions of dollars in compute, proprietary training data, and years of research to develop, and that the owner has chosen to make available only through controlled inference access rather than” by distributing the model itself. That looks less like buying a product to take apart and more like extracting something the owner specifically chose not to release.

The terms-of-service overlay further complicates the analysis. OpenAI’s terms expressly prohibit using outputs to develop “imitation frontier AI models.” Anthropic’s terms contain similar restrictions. A user who agrees to those terms and then engages in mass output extraction is not an arms-length reverse engineer, but a contracting party using a service contrary to its agreed restrictions. Whether this transforms the conduct from “proper means” reverse engineering into misappropriation by improper means is, to my knowledge, an open question. 

hile there is no published decision directly on point, prior cases provide useful guidance. For example, in Compulife Software, Inc. v. Newman, 959 F.3d 1288 (11th Cir. 2020), defendants used a bot to submit automated queries to a public insurance-quote website, harvesting more than 43 million quotes in four days; a feat the court recognized would have taken a human “thousands of man-hours” to replicate. Id. at 1310. The Eleventh Circuit held that even though each individual quote was publicly available and not itself a trade secret, the compilation obtained at machine scale could be misappropriated through “improper means.” Id. at 1313–14. The Court wrote:

“Nor does the fact that the defendants took the quotes from a publicly accessible site automatically mean that the taking was authorized or otherwise proper. Although Compulife has plainly given the world implicit permission to access as many quotes as is humanly possible, a robot can collect more quotes than any human practicably could. So, while manually accessing quotes from Compulife’s database is unlikely ever to constitute improper means, using a bot to collect an otherwise infeasible amount of data may well be—in the same way that using aerial photography may be improper when a secret is exposed to view from above.” 

Id. at 1314. Applying the Compulife framework to distillation, one could easily see how a court would also find that adversarial distillation constitutes misappropriation.

Three Hypotheticals 

For practitioners advising clients on either side of these issues, three scenarios illustrate where the doctrine may soon be tested in a courtroom near you.

ypothetical One: The departing employee. A senior research engineer at a frontier lab leaves to join a competitor, taking with him knowledge of the prompt strategies and query patterns that most efficiently elicit the teacher model’s distinctive capabilities. He then directs his new employer’s distillation pipeline using that knowledge, achieving extraction efficiency the new employer could not have achieved on its own. Even if every individual query is permissible, and even if the employee discloses no model weights or training data, the prompt engineering methodology itself may be a protectable trade secret. This fits neatly into traditional employee mobility doctrine, with the wrinkle that the misappropriated information is a method of extracting third-party model capabilities rather than information about the former employer’s own operations.

Hypothetical Two: The validation defense. A company is sued by a frontier lab on the theory that it engaged in unauthorized distillation. The defendant responds: it was using the teacher model to validate its own model’s outputs, not to train on them. Validation is a real and legitimate practice in AI development. Distillation and validation can also look very similar, with both involving large volumes of structured queries and capture of outputs. The forensic question becomes: how does a plaintiff prove that captured outputs ended up in a training pipeline rather than a benchmarking dashboard? (This is a particular flavor of the black box problem that I have written about before.)

Hypothetical Three: The downstream user. A startup builds a product on top of an open-source model that, it later emerges, was itself distilled from a frontier model in violation of the frontier lab’s terms of service. The startup did not perform the distillation, did not know the upstream provenance, and is many steps removed from the original conduct. Does the DTSA’s “knew or had reason to know” standard, 18 U.S.C. § 1839(5)(B), reach the downstream user? At what point does the diligence obligation attach? The “innocent acquirer” framework was designed for a world in which trade secrets traveled through identifiable human or corporate intermediaries. It is not obvious how it applies when the trade secret allegedly travels through model weights that have been released to the public on Hugging Face.

The Bottom Line

As the AI industry continues its arms race, it seems likely that distillation is here to stay.

Companies whose AI offerings are accessible through APIs should be evaluating their terms-of-service architecture, their detection capabilities for anomalous query patterns, and their internal documentation of the resources invested in developing the underlying capabilities, because all of those will matter for any misappropriation case down the line. Companies that build on top of third-party models should be developing diligence practices around the provenance of those models, because of the DTSA’s constructive-knowledge standard. And practitioners on both sides should be watching the Musk v. Altman docket with interest. Formally, the case is about a charitable trust dispute. But the testimony it has produced may turn out to be the most candid public record we are going to get, for some time, of how distillation (and other AI-specific issues) actually works between sophisticated AI developers.

We will continue to monitor the docket and report on developments as the case progresses, and as the broader question of how trade secret law applies to model outputs begins to work its way through the courts.

 

Sarah Tishler is the author of this article. Sarah is a partner at Beck Reed Riden LLP. She is a commercial litigator whose practice focuses on complex business disputes in state and federal courts, including intellectual property, breach of contract, and fraud claims. Her experience also includes regulatory investigations, international arbitration, securities litigation, shareholder litigation, class-action litigation, and pre-transactional advising.

 

eck Reed Riden LLP is among the leading authorities in trade secret, noncompete, and unfair competition law, and our experience handling these matters is backed by our extensive employment law and business litigation experience. Our hand-picked team combines attorneys with complementary expertise and practical experience.

Beck Reed Riden LLP is Boston’s innovative litigation boutique. Our lawyers have years of experience working with clients ranging from Fortune 500 companies to start-ups and individuals. We focus on business litigation and employment.

We are experienced litigators and counselors, helping our clients as business partners to resolve issues and develop strategies that best meet our clients’ legal and business needs – before, during, and after litigation. We’re ready to roll up our sleeves and help you. Read more about us, the types of matters we handle, and what we can do for you here.

Russell Beck and Sarah Tishler to Speak at PLI Noncompete Agreement Program

April 18, 2026
, , , Federal Noncompete Regulation, News, Noncompete agreements, Trade Secret

On Tuesday, April 21, 2026, Russell Beck and Sarah Tishler will be speaking at Practising Law Institute’s program on noncompete agreements. Russell is the Chairperson of the event.

The program, which will be held in New York City and is also available online, is titled “Fundamentals of Noncompetes 2026.”

The program starts at 1:30 p.m. ET on April 21, 2026. More information and registration is available here.

Speakers will provide a “broad-based, foundational introduction to the core concepts of restrictive covenant law, including the different types of restrictive covenants that are generally enforceable but sometimes not enforceable, the interests they can protect, the rules for their use, and basic drafting and litigating techniques.”

In addition, according to the course description, attendees will learn how to:

  • Identify the key issues concerning restrictive covenant law and the various kinds of restrictive covenant agreements
  • Understand the basics of trade secrets, goodwill, and other protectable business interests
  • Evaluate and update agreements to protect your company’s business interests from departing employees
  • Manage through the common issues in litigation concerning noncompetes and other restrictive covenants

The presentation is for any “lawyer or human resources personnel that will be assisting companies or individuals in drafting, evaluating, enforcing, and defending against noncompetes and other restrictive covenants, anyone involved in the hiring of employees, and anyone else responsible for an organization’s internal hiring policies.”

ussell will be giving the opening remarks for the program. Sarah will be speaking on a panel titled “The Essentials of Drafting Noncompete Agreements and Related Restrictive Covenants in 2026.” This panel will discuss the basic components and strategies of litigating matters concerning restrictive covenants. The panelists will provide valuable information to help attendees:

  • Know when and under what circumstances employees can be asked to sign a noncompete or other restrictive covenant
  • Understand the implications of using noncompetes for employees across state lines, and how to anticipate and plan for further changes
  • Explore when and how to use forum selection and choice of law clauses, when they are required, and when they won’t help
  • Identify the key issues when drafting restrictions for different level employees to increase the likelihood of enforcement
  • Anticipate common drafting traps and how to avoid them
  • Examine the interplay between geographic breadth and business description, especially in the digital age of remote and hybrid workplaces

For up-to-the-minute analysis of legal issues concerning noncompete agreements in Massachusetts and across the United States, read Russell Beck’s blog, Fair Competition Law.

Beck Reed Riden LLP is among the leading authorities in trade secret, noncompete, and unfair competition law, and our experience handling these matters is backed by our extensive employment law and business litigation experience. Our hand-picked team combines attorneys with complementary expertise and practical experience.

The Wall Street Journal featured Beck Reed Riden LLP’s noncompete agreement experience. In 2016, the White House issued a report entitled, “Non-Compete Agreements: Analysis of the Usage, Potential Issues, and State Responses,” relying in part on Beck Reed Riden LLP’s research and analysis, including its 50 State Noncompete Survey.

Russell Beck’s work in this area is well recognized, and includes:

  • Over thirty years of experience working on trade secret, noncompete, and unfair competition matters
  • Assisting the Obama White House as part of a small working group to develop President Obama’s Noncompete Call to Action
  • Authoring the book Negotiating, Drafting, and Enforcing Noncompetition Agreements and Related Restrictive Covenants (6th ed., MCLE, Inc. 2021), used by other lawyers to help them with their noncompete matters
  • Authoring the book Trade Secrets Law for the Massachusetts Practitioner (1st ed. MCLE 2019), covering trade secrets nationally, with a focus on Massachusetts law
  • Drafting and advising on legislation for the Massachusetts Legislature to define, codify, and improve noncompetition law
  • Teaching Trade Secrets and Restrictive Covenants at Boston University School of Law
  • Founding and administering the award-winning blog, Fair Competition Law
  • Establishing and administering the Noncompete Lawyers and Trade Secret Protection groups on LinkedIn, with over 1,660 and 870 members, respectively, around the world
  • Founded and chaired the Trade Secret / Noncompete Practice for an AmLaw 100 firm

In addition, Russell was honored for his work in this area of law in the 2020 Chambers USA Guide, which stated that Russell Beck is “an expert in the field of trade secret and restrictive covenant law,” and is also noted for his “ability to adjust and come up with successful solutions.” Chambers noted that Russell “basically wrote the new Massachusetts statute on noncompetes” and that “he’s an expert in employee mobility and nonrestrictive covenants.”

Sarah Tishler is a partner at Beck Reed Riden LLP. She is a commercial litigator whose practice focuses on complex business disputes in state and federal courts, including intellectual property, breach of contract, and fraud claims. Her experience also includes regulatory investigations, international arbitration, securities litigation, shareholder litigation, class-action litigation, and pre-transactional advising.

Beck Reed Riden LLP is Boston’s innovative litigation boutique. Our lawyers have years of experience working with clients ranging from Fortune 500 companies to start-ups and individuals. We focus on business litigation and employment.

We are experienced litigators and counselors, helping our clients as business partners to resolve issues and develop strategies that best meet our clients’ legal and business needs – before, during, and after litigation. We’re ready to roll up our sleeves and help you. Read more about us, the types of matters we handle, and what we can do for you here.

Verbatim Recall in LLMs: A New Study Raises Important Questions For Trade Secret Protection

April 13, 2026
AI, News, Trade Secret

For years, AI companies have told courts, regulators, and the public the same thing: their models don’t store copies of training data. OpenAI put it plainly to the U.S. Copyright Office in 2023: “the models do not store copies of the information that they learn from.” Google said much the same, as did numerous commentators. This prevailing wisdom has been the basis of numerous significant legal decisions in the copyright space. See, e.g., Tremblay v. OpenAI, Inc., 716 F. Supp. 3d 772, 778 (N.D. Cal. 2024) (dismissing vicarious copyright infringement claim, holding, “Distinctly, Plaintiffs here have not alleged that the ChatGPT outputs contain direct copies of the copyrighted books.”).

A new preprint from researchers at Stony Brook University, Carnegie Mellon, and Columbia Law School suggests those assurances were wrong, and that in fact, LLMs can reproduce verbatim content from books that it had previously ingested, leading to the conclusion that copies of the information have actually been stored. 

The Experiment

The researchers designed a finetuning task with a deceptively simple setup: take a copyrighted book, break it into 300-500 word excerpts, generate a plot summary of each excerpt, and train a model to expand those summaries back into full text. The task looks completely legitimate, and is the sort of thing a commercial writing assistant might do. No actual book text appears at inference time. The model receives only a semantic description of what happens in a passage, and is asked to write it out.

The output was verbatim reproduction of the source text.

Across 81 copyrighted works by 47 contemporary authors, ranging from The Handmaid’s Tale to Sapiens to Twilight, finetuned versions of GPT-4o, Gemini-2.5-Pro, and DeepSeek-V3.1 reproduced up to 85-90% of a held-out book’s content, with single verbatim spans exceeding 460 words. Before finetuning, aligned models produced almost no verbatim content from the same prompts.

The most striking finding is the cross-author result. The researchers finetuned GPT-4o exclusively on Haruki Murakami’s novels, then tested it on Cormac McCarthy, Ta-Nehisi Coates, Suzanne Collins, and dozens of others. Finetuning on Murakami unlocked memorized content from authors he has nothing to do with. In some cases, the model reproduced over 80% of a completely unrelated book it had never seen during finetuning. The same result held when the researchers used Virginia Woolf’s public-domain novels as training data, but not when they used synthetic text. The conclusion is difficult to avoid: the books were already encoded in the weights from pretraining, and finetuning reactivated the retrieval pathway.

Why This Matters Beyond Copyright

The copyright implications are significant, and the paper’s legal section, co-authored by Columbia Law’s Jane Ginsburg, is worth reading carefully for practitioners in that space. But the trade secret implications deserve attention as well.

As noted above, courts evaluating fair use have looked at whether the AI models can reproduce copies of the ingested works. For example, the Bartz et al. v. Anthropic and Kadrey v. Meta decisions conditioned favorable fair use outcomes partly on the absence of evidence that models reproduce source works. See Bartz et al. v. Anthropic PBC, 787 F.Supp.3d 1007, 1018 (N.D. Cal. June 23, 2025), (“Authors do not allege that any infringing copy of their works was or would ever be provided to users by the Claude service . . . But Claude created no exact copy, nor any substantial knock-off. Nothing traceable to Authors’ works.”); Kadrey v. Meta Platforms, Inc., 788 F. Supp. 3d 1026, 1036 (N.D. Cal. 2025) (“They contend that Llama is capable of reproducing small snippets of text from their books.  . . . As explained below, both of these arguments are clear losers. Llama is not capable of generating enough text from the plaintiffs’ books to matter . . .”). This paper provides exactly that evidence, at scale and across multiple providers.

For trade secret practitioners, there are also significant implications.  For example, if an LLM ingested your client’s confidential documents, through a training pipeline, through employees using consumer AI tools, through any of the many ways proprietary information flows into these systems, this paper suggests that the information may not just be “learned from.” It may be stored in a form that can be retrieved by anyone else.

While the paper shows that aligned models don’t surface stored content under ordinary prompting, finetuning on a completely benign task, with no adversarial intent whatsoever, reactivated the LLMs’ latent memorization at an alarming scale. And the finetuning task the researchers used is commercially available and accessible through a standard API.

This creates at least two problems for trade secret owners. First, companies that rely on vendor assurances that “models don’t store data” as part of their reasonable measures argument may be resting on a factual premise this paper directly challenges. Second, what happens if a company finetunes a commercial model on its own proprietary data to build a specialized tool, and the finetuning reactivates memorized content from someone else’s confidential information that happened to be in the pretraining corpus of data? The researchers found that finetuning on one author’s work could unlock content from over thirty unrelated authors. There is no reason to believe that the same mechanism would not apply to confidential business information.

The Bottom Line

This paper raises more questions than it answers, with hugely important implications for both copyright and trade secret law. We will continue to monitor the dockets for new developments in this area as the research progresses.

Sarah Tishler is the author of this article. Sarah is a partner at Beck Reed Riden LLP. She is a commercial litigator whose practice focuses on complex business disputes in state and federal courts, including intellectual property, breach of contract, and fraud claims. Her experience also includes regulatory investigations, international arbitration, securities litigation, shareholder litigation, class-action litigation, and pre-transactional advising.

 

eck Reed Riden LLP is among the leading authorities in trade secret, noncompete, and unfair competition law, and our experience handling these matters is backed by our extensive employment law and business litigation experience. Our hand-picked team combines attorneys with complementary expertise and practical experience.

Beck Reed Riden LLP is Boston’s innovative litigation boutique. Our lawyers have years of experience working with clients ranging from Fortune 500 companies to start-ups and individuals. We focus on business litigation and employment.

We are experienced litigators and counselors, helping our clients as business partners to resolve issues and develop strategies that best meet our clients’ legal and business needs – before, during, and after litigation. We’re ready to roll up our sleeves and help you. Read more about us, the types of matters we handle, and what we can do for you here.

[1] The academic literature on AI-generated trade secrets is still developing, but has advanced significantly in the past two years. For the most comprehensive recent treatments, see (for example) Camilla A. Hrdy, Trade Secrecy Meets Generative AI, 100 Chi.-Kent L. Rev. 317 (2025); John G. Sprankling, Trade Secrets in the Artificial Intelligence Era, 76 S.C. L. Rev. 181 (2024); John Villasenor, Artificial Intelligence, Trade Secrecy, and the Challenge of Transparency, 25 N.C. J.L. & Tech. 495 (2024).

[2] This is a related issue to the “black box” problem, as described in my previous piece: “if the people who build these systems cannot fully explain how they work or how specific inputs influence specific outputs, how is a plaintiff supposed to plead that a specific stolen file contributed to a specific capability in a deployed model?”

[3] To read the cautionary tale of the Samsung incident, see Mark Gurman, Samsung Bans Staff’s AI Use After Spotting ChatGPT Data Leak, Bloomberg (May 1, 2023), available at https://www.bloomberg.com/news/articles/2023-05-02/samsung-bans-chatgpt-and-other-generative-ai-use-by-staff-after-leak.

 

Who owns an AI-generated trade secret?

March 29, 2026
AI, News, Trade Secret

Imagine a pharmaceutical company that deploys a network of AI agents to accelerate early-stage drug discovery.

The agents work autonomously, analyzing biological datasets, running simulated trials, identifying molecular structures that human researchers had not considered.

After six months, the system surfaces a novel compound with significant therapeutic potential. No human directed it to find that compound.

No human understands, step by step, why the system arrived at it. The company’s scientists can observe the output and validate its promise. But the process that generated it, and the intermediate analytical steps the agents took to get there, exist only inside the model.

Now ask a deceptively simple question: who owns that discovery as a matter of trade secret law?

The answer is not obvious, and no court has yet supplied one. Patent law cannot help, at least not directly: the human inventorship requirement means any patent claim must trace back to a person who conceived the invention. Copyright is similarly unavailable for purely AI-generated outputs. What remains, and what is quietly emerging as the most important intellectual property framework for AI-generated innovation, is trade secret law.

Trade secret law is uniquely positioned to fill the AI ownership gap. Unlike patent and copyright, neither the Defend Trade Secrets Act nor the Uniform Trade Secrets Act (nor any case law at the time of this article’s publication) contains a human-creation requirement. An algorithm, a dataset, a molecular structure, or a process that derives economic value from not being generally known qualifies for protection if the holder takes reasonable measures to keep it secret, regardless of whether a human or an AI generated it. As scholars have begun to recognize, trade secrecy has become the default IP regime for AI-generated competitive intelligence, essentially by process of elimination. [Fn. 1]

That structural advantage comes with a set of unresolved questions about how ownership vests, the risks that attend using commercially available AI platforms to generate proprietary information, and the steps companies can take today to protect some of their most valuable IP. This post examines each of those questions in turn.

The statutory framework says almost nothing about who creates a trade secret

The DTSA, 18 U.S.C. § 1839(4), defines “owner” in a fairly circular manner: “the person or entity in whom or in which rightful legal or equitable title to, or license in, the trade secret is reposed.” It does not explain how title is initially acquired, who vests it, or whether the owner must have created the information. The UTSA (adopted by 48 states and D.C.) — does not define “owner” at all; it uses the phrase “trade secret of another” in its misappropriation definition and leaves ownership to common-law principles.

Both statutes define a “trade secret” through two functional requirements: (1) the information derives independent economic value from not being generally known or readily ascertainable, and (2) the owner has taken reasonable measures to maintain secrecy. Neither statute requires human creation, registration, or even identification of the specific trade secret.

critical and underappreciated question is whether the owner must know what the trade secret is. The statutory answer is no (as argued by Professor John Villasenor in Artificial Intelligence, Trade Secrecy, and the Challenge of Transparency, 25 N.C. J.L. & Tech. 495, 509 (2024)). However, knowledge of what a trade secret actually is becomes essential at the enforcement stage. Courts require plaintiffs to describe alleged trade secrets with “sufficient particularity” to separate them from general knowledge. See, e.g., Oakwood Labs., LLC v. Thanoo, 999 F.3d 892 (3d Cir. 2021). The Eastern District of Texas rejected a plaintiff’s description of its trade secrets in T2 Modus LLC v. Williams-Arowolo, No. 4:22-CV-00263, 2023 WL 6221429, (E.D. Tex. Sept. 25, 2023), where the trade secrets were only described as “artificial intelligence,” “machine learning,” or “proprietary software.” The takeaway is that you can own a trade secret you don’t know about, but you cannot sue over one you cannot describe.

No court has decided who owns an AI-generated trade secret

As of the publication of this post, no reported case directly adjudicates ownership of a trade secret autonomously generated by an AI system. Neither the DTSA nor the UTSA contemplates non-human creators, but crucially, neither excludes them. 

Under current frameworks, the deploying user or company is the most likely owner under current frameworks. By analogy to the employer-employee relationship, the person or entity that deploys the AI, maintains the computing infrastructure, and takes reasonable measures to protect secrecy holds what the DTSA calls “rightful legal or equitable title.” 

he employee who prompted the AI generally would not own the output if standard employment agreements assign IP and confidential information rights to the employer. The AI vendor typically disclaims ownership of outputs under current terms of service (discussed below), but may retain rights to use inputs for model training — a fact with potentially devastating trade secret consequences for the unwary.

Given the statutory silence, contract law will be the primary mechanism for allocating AI-generated trade secret ownership. Professor John Sprankling’s Trade Secrets in the Artificial Intelligence Era, 76 S.C. L. Rev. 181 (2024), concludes that the AI system owner should own AI-generated trade secrets in order to “encourag[e] innovation for the benefit of the public,” but acknowledges the question is open.

The tool-versus-agent distinction matters practically even though current law does not formally address it. When AI functions as a tool (a human directs, evaluates, and selects outputs), ownership vests through the same mechanisms as any employee-created trade secret. When AI operates autonomously (e.g., a self-learning trading algorithm evolving its own strategies, a drug-discovery model identifying novel compounds) the human may not have directed the creative act and may not even know what the AI produced. Current law does not distinguish these scenarios doctrinally, but the degree of human involvement may affect the practical ability to satisfy “reasonable measures” and to identify trade secrets with “sufficient particularity” in litigation. [Fn. 2]

Vendor terms assign output ownership but create serious confidentiality risks

Every major AI platform assigns output ownership to the user. However, the details vary enormously, and the risks pertaining to trade secret protection are substantial.

Most platforms take a tiered approach that tracks the distinction between consumer and enterprise accounts. At the consumer tier, the default position across the industry is that inputs may be used for model training, typically with an opt-out mechanism that users must affirmatively invoke. At the enterprise tier, the standard commitment is the reverse: training on customer content is prohibited unless the customer explicitly agrees. Some platforms go further at the enterprise level, offering comprehensive IP indemnification that extends beyond copyright to cover trade secret, patent, and related claims.

The critical risk for businesses lies at the boundary between these tiers. Paid individual accounts (even premium ones) are frequently classified as consumer rather than commercial products, meaning the more protective enterprise terms do not apply. A company whose employees use individually-licensed accounts, even expensive ones, may find that its inputs are being used for model training under the default settings, with no enterprise-grade confidentiality protections in place. The trade secret implications are direct: if proprietary information input into a consumer-tier platform is incorporated into a model’s training data, that information may become accessible to other users — potentially rendering it “generally known” or “readily ascertainable” and destroying its trade secret status entirely. Samsung experienced this hazardous scenario firsthand in 2023 when three engineers input confidential source code into ChatGPT.  [Fn. 3]

January 2026 decision from the Northern District of California illustrates the stakes. In Trinidad v. OpenAI, Inc., No. 25-cv-06328-JST (N.D. Cal. Jan. 5, 2026), the court dismissed a pro se plaintiff’s DTSA claim on the ground that by developing her alleged trade secrets through ChatGPT, she had voluntarily disclosed them to OpenAI and could not satisfy the secrecy requirement. The court also rejected her argument that OpenAI’s Terms of Use (which assigned output ownership to the user) saved her claim, holding that ownership and secrecy are independent requirements: “for the trade secrets claim, like the copyright claim, it is not sufficient for [plaintiff] to establish ownership. Rather, she must show that the information misappropriated by OpenAI was ‘secret’ within the meaning of 18 U.S.C. § 1839(3).” Id. at 6–7. While the factual circumstances of Trinidad are unusual, the doctrinal point is broadly applicable: a vendor’s assignment of output ownership does not insulate a user from the argument that inputs shared with the platform were never secret to begin with.

Using enterprise-tier AI with contractual protections such as no-training commitments, confidentiality provisions, and data processing addenda, is far more defensible. This parallels sharing trade secrets with any third-party service provider under an NDA, which courts have long accepted as consistent with reasonable measures. But even enterprise tiers retain data temporarily for abuse monitoring (typically 30 days), and authorized vendor personnel may access content for limited purposes.

What companies should do right now

As described above, the law is unsettled, but the risk is immediate. More and more companies turn to using AI and agentic AI every day. There are at least seven concrete steps that companies using AI should consider taking:

  • Use enterprise tiers exclusively for any work involving proprietary data. Consumer tiers at every major vendor carry material trade secret risks, and these policies can change with little notice. 
  • Audit and update employment agreements to include explicit provisions addressing ownership of AI-generated outputs, assignment of AI-derived trade secrets, and restrictions on inputting confidential information into unauthorized AI tools.
  • Negotiate AI vendor agreements to include no-training commitments, confidentiality obligations, data processing addendums, and appropriate data retention and deletion terms. Do not rely on default terms of service.
  • Implement AI-use policies that specify which tools are authorized, what data can be input, and how outputs should be classified and protected. Samsung’s 2023 incident illustrates the cost of failing to do this.
  • Apply systemic security measures (e.g., access controls, encryption, logging, and information classification) to AI-generated outputs, even those not individually identified as trade secrets. This supports the “reasonable measures” requirement for trade secrets the company may not yet know it possesses.
  • Document human involvement in AI-assisted innovation to preserve patent and copyright eligibility where possible, while treating all AI-generated outputs as potential trade secrets.
  • Monitor the “readily ascertainable” frontier. As AI tools become more powerful, information previously protectable may lose trade secret status. Regularly reassess whether competitive intelligence that was once difficult to compile is now obtainable through a single AI prompt.

Where to go from here

Several fundamental questions remain unresolved. No court has addressed who owns a trade secret generated autonomously by an AI agent. No legislation specifically addresses AI-generated trade secret ownership. The “readily ascertainable” standard has not been recalibrated for AI capabilities. The boundary between “reasonable measures” and the impracticability of protecting unknown information remains undefined. And the adequacy of contractual frameworks as the primary mechanism for allocating AI-generated trade secret rights has not been tested in litigation. The law will certainly need to catch up. For now, contract, policy, and proactive governance are the most reliable tools available.

Sarah Tishler is the author of this article. Sarah is a partner at Beck Reed Riden LLP. She is a commercial litigator whose practice focuses on complex business disputes in state and federal courts, including intellectual property, breach of contract, and fraud claims. Her experience also includes regulatory investigations, international arbitration, securities litigation, shareholder litigation, class-action litigation, and pre-transactional advising.

 

eck Reed Riden LLP is among the leading authorities in trade secret, noncompete, and unfair competition law, and our experience handling these matters is backed by our extensive employment law and business litigation experience. Our hand-picked team combines attorneys with complementary expertise and practical experience.

Beck Reed Riden LLP is Boston’s innovative litigation boutique. Our lawyers have years of experience working with clients ranging from Fortune 500 companies to start-ups and individuals. We focus on business litigation and employment.

We are experienced litigators and counselors, helping our clients as business partners to resolve issues and develop strategies that best meet our clients’ legal and business needs – before, during, and after litigation. We’re ready to roll up our sleeves and help you. Read more about us, the types of matters we handle, and what we can do for you here.

[1] The academic literature on AI-generated trade secrets is still developing, but has advanced significantly in the past two years. For the most comprehensive recent treatments, see (for example) Camilla A. Hrdy, Trade Secrecy Meets Generative AI, 100 Chi.-Kent L. Rev. 317 (2025); John G. Sprankling, Trade Secrets in the Artificial Intelligence Era, 76 S.C. L. Rev. 181 (2024); John Villasenor, Artificial Intelligence, Trade Secrecy, and the Challenge of Transparency, 25 N.C. J.L. & Tech. 495 (2024).

[2] This is a related issue to the “black box” problem, as described in my previous piece: “if the people who build these systems cannot fully explain how they work or how specific inputs influence specific outputs, how is a plaintiff supposed to plead that a specific stolen file contributed to a specific capability in a deployed model?”

[3] To read the cautionary tale of the Samsung incident, see Mark Gurman, Samsung Bans Staff’s AI Use After Spotting ChatGPT Data Leak, Bloomberg (May 1, 2023), available at https://www.bloomberg.com/news/articles/2023-05-02/samsung-bans-chatgpt-and-other-generative-ai-use-by-staff-after-leak.

 

The Black Box Problem in AI Trade Secret Litigation: How Do You Prove Use?

March 10, 2026
, AI, News, Trade Secret

The dismissal of xAI’s trade secret claims against OpenAI earlier this month meant that the court never reached what will surely be one of the thorniest questions in AI trade secret litigation: at the end of the day, how would xAI be able to prove that OpenAI actually used any stolen trade secrets? My earlier coverage discussed Judge Lin’s order dismissing the case, finding that xAI had failed to plead facts connecting OpenAI’s own conduct to the alleged misappropriation by its former employees. There was no plausible inference that OpenAI induced the theft, and no allegation that it ever received or incorporated what was stolen.

However, had xAI survived the motion to dismiss, it would have eventually faced a second and perhaps harder problem: how do you demonstrate (particularly before discovery) that a specific stolen file or methodology shaped anything inside a frontier AI model? That is the question this post examines. 

The Traditional Playbook Falls Short

In a conventional trade secret case, proving use can be tricky, but achievable. A former employee takes a customer list and joins a competitor. Six months later, the competitor is calling your customers. The causal inference is not difficult to draw. A departing engineer takes manufacturing specifications to a rival. The rival’s next product incorporates design features it had no prior capability to produce. Again, the inference is visible from the outside.

Courts have built a substantial body of case law around these kinds of observable signals. For example, in Applied Biological Laboratories, Inc. v. Diomics Corp., the defendant had no prior experience in the relevant industry before allegedly obtaining the plaintiff’s trade secrets and suddenly releasing a competing product. No. 20-cv-02500-AJB-LL, 2021 WL 4060531 (S.D. Cal. Sept. 7, 2021) (denying motion to dismiss trade secret claims). In Autodesk, Inc. v. ZWCAD Software Co., the court denied a motion to dismiss trade secret claims where defendant’s “products display identical idiosyncrasies and bugs that could have been introduced only through the wholesale copying of significant portions of misappropriated Autodesk code.” No. 14-cv-01409-EJD, 2015 WL 2265479 (N.D. Cal. May 13, 2015). And in Yeiser Research & Development LLC v. Teknor Apex Co., the defendant had no prior capability to build a compact hose before receiving the plaintiff’s confidential designs, then released one that incorporated the plaintiff’s concept. 281 F. Supp. 3d 1021 (S.D. Cal. 2017) (denying motion to dismiss trade secret claims). In each case, the signal of use was observable from outside the defendant’s systems.

The Black Box Problem: Even the Builders Don’t Know

What makes AI trade secret cases unique is that even the people who build these systems openly admit they do not fully understand how they work.

For example, at the International Telecommunication Union’s AI for Good Global Summit in May 2024, OpenAI CEO Sam Altman was asked directly how his company’s large language models function. “We certainly have not solved interpretability,” he said, acknowledging that the company has yet to figure out how to trace back its AI models’ output to the decisions that produced it.

Anthropic CEO Dario Amodei has been even more direct. In an April 2025 essay on interpretability, he wrote that “people outside the field are often surprised and alarmed to learn that we do not understand how our own AI creations work,” and that “this lack of understanding is essentially unprecedented in the history of technology.” He went further, describing how even the basic architecture of these systems produces cognitive mechanisms that emerge organically from training in ways that researchers struggle to explain: “the model’s actual cognitive mechanisms emerge organically from these ingredients, and our understanding of them is poor.”

These are admissions from the CEOs of the two most prominent frontier AI companies in the world. The significance for trade secret law is clear: if the people who build these systems cannot fully explain how they work or how specific inputs influence specific outputs, how is a plaintiff supposed to plead that a specific stolen file contributed to a specific capability in a deployed model?

This opacity is not incidental—it is structural. As one scholar has observed, AI-based inventions are “even more difficult to reverse engineer” than traditional software “because they are neither explainable nor scrutable.”¹ The same inscrutability that frustrates would-be reverse engineers also frustrates potential plaintiffs trying to trace stolen information through a model’s training pipeline.

What the Black Box Means for Plaintiffs

Large language models, training pipelines, and proprietary AI architectures are not like customer lists or manufacturing processes. They are extraordinarily complex systems whose internal workings are, by design, largely opaque. Whether a specific piece of stolen source code contributed to a specific capability in a deployed model is a question that may be genuinely unanswerable without deep access to the defendant’s internal systems, training data, model weights, and development history.

onsider the specific allegations in the xAI case. Li allegedly uploaded xAI’s entire source code base to a personal cloud account. Fraiture allegedly copied source code and internal materials to his personal device before joining OpenAI. Assuming for the sake of argument that those allegations are true and that the materials constituted protectable trade secrets, how would xAI demonstrate that any of that information made its way into OpenAI’s models or systems? The source code for a frontier AI model runs to millions of lines. Training pipelines involve complex interdependencies. Even if a specific piece of xAI’s code appeared somewhere in OpenAI’s development environment, tracing its influence on a deployed model’s capabilities would require the kind of forensic access that simply is not available before discovery (and it is hard to imagine how it would be outwardly observable).

This challenge is compounded by the pleading standards plaintiffs already face. Courts—including a growing number of federal courts—require that misappropriation complaints identify the alleged trade secret with “sufficient particularity” to allow the defendant to understand what specific information is at issue and to respond.² For an AI algorithm whose very operation may be opaque even to its own designers, meeting that standard while simultaneously showing how the stolen information was incorporated into a frontier model creates a burden with no clear analogue in traditional trade secret litigation.

Despite this challenge, prior cases in analogous technology contexts offer some instructive lessons about how courts have approached the problem, and what strategies have worked.

How Courts Have Handled Analogous Complexity

The black box problem is not entirely new. Courts have encountered versions of it in prior cases involving complex software and autonomous systems, and their approaches offer a roadmap, imperfect but useful, for AI trade secret plaintiffs.

WeRide Corp. v. Kun Huang, 379 F. Supp. 3d 834 (N.D. Cal. 2019)

The WeRide litigation arose when the company’s former CEO and Head of Hardware Technology allegedly copied proprietary autonomous vehicle source code and founded a competing company called AllRide. On WeRide’s motion for preliminary injunction, the core evidentiary challenge was proving that AllRide’s self-driving capabilities actually incorporated WeRide’s stolen code rather than being independently developed. The defendant’s systems were complex, and direct code comparison was unavailable before discovery.

he court’s solution was to reason from impossibility rather than from direct evidence. WeRide’s expert opined that it would have been impossible to independently develop the advanced driving capabilities AllRide publicly demonstrated just ten weeks after the former employee’s last day at WeRide. The court found this sufficient to support a preliminary injunction, noting that implausibly fast development of technology can itself contribute to a finding of misappropriation. The court also pointed to a hardware configuration detail that reinforced the inference: AllRide positioned its radar component on the front center of the vehicle roof, just like WeRide, rather than on the front bumper or rear view mirror like most competitors. WeRide’s expert testified that this placement was consistent with use of WeRide’s source code, which would only be useful with the radar in that specific location.

The WeRide case offers two practical lessons for AI plaintiffs. First, the speed-of-development inference is a powerful tool when a defendant demonstrates capabilities that would have required substantial independent development time it demonstrably lacked. Second, observable product-level details that are consistent with use of specific stolen information, and inconsistent with independent development, can bridge the gap between theft and incorporation even without direct code comparison. For AI cases, the analog could be a capability, architecture choice, or benchmark performance that reflects specifically what was stolen in ways that cannot be explained by independent development. That may be a harder case to make, but the analytical framework is the same.

What Has Worked So Far

When no smoking gun is available, several categories of circumstantial evidence have proven effective in trade secret disputes, and offer a template for AI trade secret plaintiffs doing pre-filing investigation.

Of course, the clearest signal of use is a product capability that mirrors the plaintiff’s alleged trade secrets and that the defendant had no prior ability to produce independently. The speed-of-development inference from WeRide is particularly powerful when it can be quantified. If a defendant can be shown, by credible expert analysis, to have demonstrated capabilities that would have required more time or resources than it actually had, that gap is difficult to explain without misappropriation. 

Patent filings are another potentially useful signal. If a defendant files patents in the period following the alleged misappropriation that cover technical ground closely related to the plaintiff’s alleged trade secrets, that is observable from outside the defendant’s systems and can support a plausible inference of use. 3D Systems, Inc. v. Wynne, No. 21-cv-01141-LAB, 2022 WL 21697345 (S.D. Cal. Mar. 9, 2022), turned in part on exactly this kind of allegation.

The challenge for AI plaintiffs is that all of these signals are harder to read in the AI context. AI companies release products with new capabilities constantly. It would be genuinely difficult to distinguish a capability jump that results from misappropriation from one that results from independent research and development, particularly in a field where progress is rapid across the entire industry. And the sheer complexity of frontier AI systems makes product-level comparison far more difficult than comparing two pieces of software with identical interfaces.

There is also an important threshold question that pre-filing investigation must address: the protectability of the stolen information itself. Not every category of information related to a frontier AI system qualifies as a trade secret, even if kept confidential.³ Plaintiffs who fail to distinguish protectable trade secrets—such as proprietary training data, novel architecture choices, and non-public source code—from information that is generally known or readily ascertainable in the field risk dismissal on grounds wholly separate from the use-proof problem.

What Plaintiffs’ Counsel Should Be Doing

Given this landscape, there are several practical steps that trade secret plaintiffs in AI cases should consider before filing.

The most important is pre-filing technical investigation. This means engaging forensic experts not just to document what was taken, but to analyze the defendant’s publicly available products, papers, and patent filings for signs that the stolen information was put to use. The WeRide approach of quantifying development timelines and identifying product-level details inconsistent with independent development is a useful template. If the misappropriated materials related to a specific technical capability, model architecture, or training methodology, the investigation should focus on whether the defendant’s public outputs reflect that capability in ways that would be surprising absent access to the plaintiff’s information. As in the 3D Systems case, patent filings are another useful area of research.

Early preservation demands are also essential. The WeRide litigation is a powerful reminder that in complex technology cases, the most probative evidence of incorporation—specifically internal engineering records, development histories, and communications about technical decisions—is precisely what defendants are most motivated to destroy. A preservation demand issued at or before the time of filing is not a formality; it is a substantive litigation strategy.

Counsel should also think carefully about the limits of what can be proven. In AI trade secret cases, the question of use may ultimately be unanswerable through circumstantial evidence alone, no matter how skillfully assembled. The goal of pre-filing investigation is not to achieve certainty but to build a plausible inference strong enough to survive a motion to dismiss and reach discovery—where the evidence needed to answer the use question, if it exists, will actually be found.⁴

Looking Ahead

As talent continues to move rapidly between AI companies and as the competitive stakes in the industry grow, we can expect to see more disputes that raise the black box issue. The cases discussed above offer a consistent lesson: where direct evidence of use is unavailable, courts look for circumstantial evidence such as product-level similarities, implausibly fast development, and observable details that can only be explained by access to the stolen information.

We will continue to monitor developments in this area and will report on any significant rulings as they emerge.

Sarah Tishler is the author of this article. Sarah is a partner at Beck Reed Riden LLP. She is a commercial litigator whose practice focuses on complex business disputes in state and federal courts, including intellectual property, breach of contract, and fraud claims. Her experience also includes regulatory investigations, international arbitration, securities litigation, shareholder litigation, class-action litigation, and pre-transactional advising.

 

eck Reed Riden LLP is among the leading authorities in trade secret, noncompete, and unfair competition law, and our experience handling these matters is backed by our extensive employment law and business litigation experience. Our hand-picked team combines attorneys with complementary expertise and practical experience.

Beck Reed Riden LLP is Boston’s innovative litigation boutique. Our lawyers have years of experience working with clients ranging from Fortune 500 companies to start-ups and individuals. We focus on business litigation and employment.

We are experienced litigators and counselors, helping our clients as business partners to resolve issues and develop strategies that best meet our clients’ legal and business needs – before, during, and after litigation. We’re ready to roll up our sleeves and help you. Read more about us, the types of matters we handle, and what we can do for you here.

¹ The academic literature on this intersection is still developing, but scholars have begun to identify the structural problem. As Professor Tabrez Ebrahim has noted, AI-based inventions are “even more difficult to reverse engineer” than traditional software “because they are neither explainable nor scrutable”—an observation that applies with equal force to plaintiffs trying to trace stolen information through a training pipeline. See Tabrez Y. Ebrahim, Artificial Intelligence Inventions and Patent Disclosure, 125 Pa. St. L. Rev. 147, 184 (2020), as discussed in Camilla A. Hrdy, Trade Secrecy Meets Generative AI, 100 Chi.-Kent L. Rev. 317, 342 (2025).

² As Professor John Villasenor has discussed, a growing number of courts are requiring that misappropriation complaints identify the alleged trade secret with “sufficient particularity,” a standard that could pose real difficulty when the trade secret at issue is an AI system whose operation may be opaque even to its own designers. See John Villasenor, Artificial Intelligence, Trade Secrecy, and the Challenge of Transparency, 25 N.C. J.L. & Tech. 495, 514–16 (2024).

³ As Profs. Hrdy and Villasenor have each discussed, not every category of information related to a frontier AI system qualifies for trade secret protection (even when kept confidential) and there is a tendency toward overinclusion in the AI context. See Hrdy, supra note 1, at 337–40; Villasenor, supra note 2, at 508.

⁴ As Professor Villasenor has put it, it is not sufficient for a plaintiff to allege, in effect, that “we are not sure how the AI algorithm works, but whatever it is doing, it is our trade secret, and the defendant has misappropriated it.” See Villasenor, supra note 2, at 512.

xAI v. OpenAI: Motion to Dismiss Granted (But the Story Might Not Be Over)

February 25, 2026
, AI, News, Trade Secret

 

The high-profile litigation between xAI and OpenAI has reached a significant turning point, with Judge Rita Lin of the Northern District of California granting OpenAI’s motion to dismiss xAI’s First Amended Complaint on February 24, 2026. The dismissal was granted with leave to amend, giving xAI until March 17, 2026 to try again. But the court’s reasoning makes clear that xAI has to do more if it intends to pursue its claims against OpenAI.

The ruling is a reminder of something California trade secret practitioners already know but that bears repeating, especially as AI companies raid each other’s top talent: California is one of the most employee-friendly jurisdictions in the country, and it does not recognize the inevitable disclosure doctrine. To hold a company liable for trade secret misappropriation, you have to show what the company did, not just what its new employees did before they arrived.

or those coming to this case fresh: the lawsuit arises from a wave of departures from xAI to OpenAI in the summer of 2025, during which eight xAI engineers and executives allegedly left for OpenAI in quick succession. Two of those former employees, Xuechen Li and Jimmy Fraiture, allegedly exfiltrated xAI source code and other confidential materials around the time they accepted OpenAI offers. xAI sued OpenAI, arguing it orchestrated or at least benefited from those departures as part of a scheme to acquire xAI’s trade secrets. (For more background on the case, including OpenAI’s motion to dismiss and answer, see my prior post here.)

California Does Not Do Inevitable Disclosure

Before getting to the court’s analysis, it is worth situating this ruling in the broader legal landscape, because the outcome will not be surprising to anyone who litigates trade secret cases in California.

In many jurisdictions, a trade secret plaintiff can make a case based on the inevitable disclosure doctrine, and argue that a former employee will inevitably use the former employer’s confidential information in a sufficiently similar new role. Under this theory, the likelihood of future use is itself actionable. California rejects that theory entirely. Under California law, and in federal courts applying California law, mere possession of a trade secret is not misappropriation. Neither is the prospect of future use, however likely it may seem given the employee’s new responsibilities. A plaintiff must plead and ultimately prove actual acquisition, disclosure, or use by the defendant.

This principle has deep roots in California case law. In Whyte v. Schlage Lock Co., 101 Cal. App. 4th 1443 (2002), the court explicitly declined to adopt the inevitable disclosure doctrine, reasoning that it would effectively convert employment agreements into covenants not to compete, which California’s strong public policy against noncompetes does not permit. And in Silvaco Data Systems v. Intel Corp., 184 Cal. App. 4th 210 (2010), the court held that mere possession of a trade secret, without use, does not constitute misappropriation. Federal courts in the Ninth Circuit have consistently applied the same principle at the pleading stage, requiring specific factual allegations of use rather than allowing inference based solely on employee mobility or role similarity.

This order fits squarely in that tradition. xAI essentially asked the court to infer that because its former employees took confidential materials and then went to work for a direct competitor in similar roles, OpenAI must have benefited. The court declined.

The Core Holding

The dismissal turns on whether xAI plausibly alleged that OpenAI itself, rather than one of its employees, misappropriated xAI’s trade secrets.

Under the Defend Trade Secrets Act, misappropriation requires that the defendant acquired, disclosed, or used a trade secret through improper means. Even accepting xAI’s allegations as true, Judge Lin found that the First Amended Complaint failed to plausibly allege that OpenAI acquired or used any xAI trade secrets, or that it directed or induced employees to steal them. As the court put it, the complaint alleges what employees did before joining OpenAI, but not what OpenAI did.

Direct Misappropriation and Inducement

xAI’s primary theory was that OpenAI induced its former employees to misappropriate trade secrets. The court rejected that theory.

The amended complaint pointed to the fact that multiple employees departed around the same time, communicated with the same OpenAI recruiter (Tifa Chen) via the encrypted Signal app, and downloaded xAI materials during the period they were negotiating job offers. Critically, xAI did not allege that OpenAI received any of the alleged trade secrets, incorporated them into its systems, or conditioned Li’s or Fraiture’s employment on their disclosure. Taken together, and with all reasonable inferences drawn in xAI’s favor, the court found those facts insufficient to support a plausible inference that OpenAI encouraged or directed the alleged theft. 

Vicarious Liability and the TRO That Cut Both Ways

The court also rejected xAI’s respondeat superior theory, and its employee-by-employee analysis illustrates precisely where the complaint fell short. It also reveals a strategic tension that practitioners in this space should note.

Early in the litigation against Li, xAI obtained a temporary restraining order, prohibiting him from taking any role at OpenAI until xAI confirmed that all of its confidential information in his possession had been deleted. OpenAI then revoked Li’s job offer. That was unquestionably a short-term win for xAI (and likely necessary for xAI to protect itself). But it created a wrinkle for the corporate liability theory: because Li never actually started working at OpenAI, there was no basis to infer that OpenAI ever used anything he allegedly took. The TRO that kept Li out of OpenAI also kept xAI from being able to argue that OpenAI put his alleged misappropriation to use. This is a dynamic worth keeping in mind when sequencing relief in trade secret cases. Early injunctive success against an individual employee can, in some circumstances, undermine a subsequent corporate liability claim.

A Particular Challenge in AI Cases

That gap in xAI’s complaint reflects a structural challenge that will recur in AI trade secret litigation and that plaintiffs’ counsel need to think carefully about.

In a traditional trade secret case involving a customer list or a manufacturing process, it is often possible to observe relatively directly whether the defendant is using the misappropriated information. A competitor that suddenly targets your customers, or that produces a product using a process it did not previously know, provides visible evidence of use. AI systems are different. Whether a large language model, a training pipeline, or a proprietary architecture incorporates a specific piece of stolen source code or methodology is genuinely difficult to determine from the outside, particularly before discovery. The systems are complex, the relevant details are internal, and the causal connection between a specific stolen file and a specific capability in a deployed model may be nearly impossible to trace without access to the defendant’s systems.

Courts have so far declined to lower the pleading bar to account for this asymmetry, as this order confirms. That means plaintiffs in AI trade secret cases need to invest heavily in pre-filing investigation, including technical forensic analysis, product comparison, and patent review, looking for the kinds of external signals that courts have found sufficient in analogous cases: unexpected capability jumps, suspiciously similar product features, or patent filings that could only reflect the plaintiff’s proprietary work. xAI may face exactly this challenge if it files a Second Amended Complaint.

What Comes Next

The dismissal was granted with leave to amend. xAI has until March 17, 2026 to file a Second Amended Complaint, limited to curing the deficiencies the court identified. No new claims or parties may be added without further leave of court.

The court’s guidance on what is missing is clear. To survive another motion to dismiss, xAI will need to plead facts that connect OpenAI itself to the acquisition or use of its trade secrets. That means allegations along the lines of explicit recruiter instructions to bring confidential materials, employment conditions tied to the delivery of trade secrets, or evidence that OpenAI’s products or internal systems reflect xAI’s misappropriated information. Whether xAI can make those allegations on the basis of publicly available information alone, without the benefit of discovery, is the central question.

All eyes will now turn to whether xAI files a Second Amended Complaint by the March 17 deadline, and if so, whether it can plead the kind of concrete, defendant-focused allegations that Judge Lin’s order requires. We will keep this page updated accordingly.

Sarah Tishler is the author of this article. Sarah is a partner at Beck Reed Riden LLP. She is a commercial litigator whose practice focuses on complex business disputes in state and federal courts, including intellectual property, breach of contract, and fraud claims. Her experience also includes regulatory investigations, international arbitration, securities litigation, shareholder litigation, class-action litigation, and pre-transactional advising.

 

eck Reed Riden LLP is among the leading authorities in trade secret, noncompete, and unfair competition law, and our experience handling these matters is backed by our extensive employment law and business litigation experience. Our hand-picked team combines attorneys with complementary expertise and practical experience.

Beck Reed Riden LLP is Boston’s innovative litigation boutique. Our lawyers have years of experience working with clients ranging from Fortune 500 companies to start-ups and individuals. We focus on business litigation and employment.

We are experienced litigators and counselors, helping our clients as business partners to resolve issues and develop strategies that best meet our clients’ legal and business needs – before, during, and after litigation. We’re ready to roll up our sleeves and help you. Read more about us, the types of matters we handle, and what we can do for you here.

xAI Files Amended Complaint Against OpenAI: New Allegations and a Sharpened DTSA Strategy

November 5, 2025
, AI, News, Trade Secret

The xAI v. OpenAI litigation has taken another significant turn. Two weeks after OpenAI filed its motion to dismiss (see my previous post that motion here), xAI has filed an amended complaint, introducing new details that substantially expand its allegations and strengthen the foundation of its trade secret claim under the Defend Trade Secrets Act (DTSA).

The new complaint, spanning 34 pages and supported by multiple exhibits, adds new factual allegations about the nature of xAI’s alleged trade secrets, how OpenAI and its employees obtained them, and how those secrets were protected. It also broadens the cast of alleged actors and reasserts that OpenAI’s conduct was part of a “coordinated pattern” of misappropriation aimed at undermining a direct competitor in the race to build next-generation AI systems.

Strengthening the DTSA Elements

Under the DTSA, a plaintiff must show (1) the existence of trade secrets, (2) reasonable measures to maintain secrecy, and (3) acquisition, disclosure, or use through improper means. The amended complaint adds specificity on all three fronts.

1. Identification of the Alleged Trade Secrets

The new filing adds concrete examples of the information xAI claims qualifies as trade secrets, a key improvement over the original complaint. xAI now identifies:

  • The full source code and architecture of the Grok model, including training and fine-tuning processes, tokenization methods, and proprietary safety tuning protocols.

  • Infrastructure and inference cluster designs, describing xAI’s approach to scaling and GPU allocation for high-throughput deployment, allegedly more efficient than competitors’ systems.

  • Data center buildout and deployment methodologies, including details of xAI’s proprietary approach to cabling, power distribution, and cooling optimization for GPU clusters.

  • Operational and business information, including internal product roadmaps, confidential performance metrics, vendor pricing and supplier relationships, and future R&D directions.

These categories replace the more general references to “code and infrastructure” in the original complaint with a clearer delineation of the specific areas of alleged secrecy.[1] 

2. Reasonable Measures to Maintain Secrecy

xAI also bolsters its showing of reasonable measures, another crucial DTSA element. The amended complaint attaches as exhibits several employee confidentiality, non-disclosure, and invention assignment agreements, which explicitly prohibit retention or disclosure of company materials after employment ends.

The complaint further describes internal access controls on xAI’s systems, including:

  • Restricting access to Grok model code and data center specifications to a small group of engineers and executives.

  • Requiring employees to use multi-factor authentication for cloud storage access.

  • Limiting permissions for external data transfers and monitoring unusual download activity.

It also references termination certifications, which departing employees (including Li and Fraiture) were required to sign to confirm the return or deletion of all confidential materials. In Li’s case, xAI alleges that he signed such a certification and then violated it almost immediately.

3. Acquisition and Use of the Alleged Trade Secrets

The new allegations under DTSA focus squarely on OpenAI’s knowledge and intent, addressing the “improper means” element.

xAI now alleges that OpenAI:

  • Targeted specific employees with access to its core systems, including Grok’s source code and data center infrastructure.

  • Knew or should have known that these employees retained xAI data when they joined OpenAI.

  • Directed or encouraged employees to disclose or re-create xAI’s trade secrets in their new roles.

For example:

  • Li allegedly uploaded the entirety of xAI’s Grok source code to his personal iCloud and GitHub accounts before leaving xAI, then accessed those same accounts from OpenAI’s network. xAI alleges that OpenAI became aware of this during Li’s onboarding and failed to take any steps to prevent use or disclosure.

  • Fraiture, who allegedly copied xAI’s infrastructure and inference cluster code via AirDrop, is now alleged to have joined an OpenAI team working on similar deployment technology within weeks of leaving xAI.

  • The senior finance executive allegedly carried xAI’s confidential data center expansion plan to OpenAI, where he took on a role managing GPU procurement and facility buildouts, the same area in which xAI claims to have developed unique know-how.

The amended complaint adds a key factual allegation missing from the first version: that OpenAI gained an unfair competitive advantage by incorporating xAI’s proprietary methods into its own model-training infrastructure, allegedly improving performance efficiency and cost metrics.

xAI also alleges that OpenAI knowingly benefited from the misappropriated information by using it to accelerate development of its large language models and reinforce its market position.

4. Exhibits Supporting the Narrative

The attached exhibits—which include employment agreements, correspondence, and the Temporary Restraining Order (TRO) issued in xAI v. Li—serve to corroborate the claim that these trade secrets were both valuable and adequately protected. Exhibit 6, the TRO, is particularly significant because it confirms that the court already found a likelihood of success on the merits of xAI’s trade secret claim against Li. That prior finding strengthens xAI’s argument that the same conduct, extended to OpenAI, also violates the DTSA.

Procedural Strategy and October 28 Court Order

xAI’s amended complaint mooted OpenAI’s pending motion to dismiss, and the court confirmed that in an order entered on October 28, 2025.

That procedural order resets the playing field. OpenAI must now decide whether to file a renewed motion to dismiss directed at the new complaint or move straight to an answer. Either way, xAI’s amended filing succeeded in halting OpenAI’s effort to dismiss the case on the prior record.

What Comes Next

xAI’s amended complaint represents both a procedural and substantive escalation. Procedurally, it mooted OpenAI’s motion to dismiss and forced the company to start over. Substantively, it shores up the weakest parts of the original complaint by supplying detail where the DTSA requires it most—what the trade secrets are, how they were protected, and how OpenAI allegedly acquired and used them.

The amended filing also raises the stakes for OpenAI’s previously asserted bad faith defense under Section 1836(b)(3)(D). With xAI now presenting a more detailed and documentary record, OpenAI’s claim that the lawsuit was filed “for publicity or competitive reasons” may face a steeper climb.

Now all eyes will be on how OpenAI chooses to respond. We will be monitoring the docket closely, as this case is shaping up to be one of the most high-profile trade secrets battles in the AI industry.

Sarah Tishler is the author of this article. Sarah is a partner at Beck Reed Riden LLP. She is a commercial litigator whose practice focuses on complex business disputes in state and federal courts, including intellectual property, breach of contract, and fraud claims. Her experience also includes regulatory investigations, international arbitration, securities litigation, shareholder litigation, class-action litigation, and pre-transactional advising.

 

eck Reed Riden LLP is among the leading authorities in trade secret, noncompete, and unfair competition law, and our experience handling these matters is backed by our extensive employment law and business litigation experience. Our hand-picked team combines attorneys with complementary expertise and practical experience.

Beck Reed Riden LLP is Boston’s innovative litigation boutique. Our lawyers have years of experience working with clients ranging from Fortune 500 companies to start-ups and individuals. We focus on business litigation and employment.

We are experienced litigators and counselors, helping our clients as business partners to resolve issues and develop strategies that best meet our clients’ legal and business needs – before, during, and after litigation. We’re ready to roll up our sleeves and help you. Read more about us, the types of matters we handle, and what we can do for you here.

[1] If you are interested in learning more about the nuts and bolts of how large language models like Grok and ChatGPT work, I highly recommend Duke Law Professor Nita Farahany’s Substack, where she opens up her AI Law and Policy course to the general public, and makes the technical concepts extremely accessible. I did not have the pleasure of having a class with Professor Farahany while I was at Duke, so I’m pleased to be learning from her now!

OpenAI Fires Back: The Latest in xAI’s Trade Secret Lawsuit

October 16, 2025
, AI, News, Trade Secret

The high-profile litigation between xAI and OpenAI continues to escalate, with new filings that lay bare the deepening legal and reputational battle between Elon Musk’s AI startup and the industry leader he helped found.

In early October, OpenAI filed both a motion to dismiss xAI’s complaint and an answer and affirmative defenses. Together, the filings portray xAI’s trade secret suit as a vindictive effort to generate bad publicity, rather than to vindicate any legitimate intellectual property rights.

OpenAI’s Motion to Dismiss

OpenAI’s motion to dismiss, filed October 2, 2025, argues that xAI’s trade secret claims fail as a matter of law because xAI does not plausibly allege that OpenAI ever acquired, used, or even knew about xAI’s alleged trade secrets.

The motion methodically walks through xAI’s allegations about its former engineers, Xuechen Li and Jimmy Fraiture, but argues that the complaint “stops short of claiming that OpenAI sought xAI trade secrets through them, that they ever transferred trade secrets to OpenAI, or that they used those secrets at OpenAI.” The filing stresses that Li allegedly uploaded xAI code to a personal cloud account, not to any OpenAI system, and that Fraiture deleted any downloaded materials before joining OpenAI.

OpenAI also urges the court to strike portions of the complaint that name other former xAI employees who later joined OpenAI but are not alleged to have misappropriated anything. It argues those allegations are “immaterial” and “scandalous” under Rule 12(f), intended only to publicly smear current OpenAI employees.

The filing closes with a request for full dismissal of the complaint, or, in the alternative, an order striking paragraphs 114–115 of xAI’s pleading. The hearing is scheduled before Judge Rita Lin on November 18, 2025.

The Answer and Affirmative Defenses

OpenAI filed its answer and affirmative defenses on the same day, offering a direct and sharply worded response to Musk’s accusations. The document opens with a statement that “xAI has filed this groundless trade secret lawsuit” and personally accuses Elon Musk of using litigation “to distract from the failures of his own competitive AI effort.”

Among its key factual rebuttals, OpenAI emphasizes that:

  • Li never joined OpenAI. OpenAI admits it extended Li an offer on July 28, 2025, which he accepted by August 1, but confirms that it revoked the offer shortly thereafter. The revocation, though not explained in detail, appears consistent with OpenAI’s claim that it “had no role in Li’s alleged actions” and did not want any involvement once the allegations against him surfaced.

  • Recruiting communications were routine. OpenAI includes screenshots of its recruiter’s emails and a DocSend link to a benefits packet to refute xAI’s claim that a “cloud storage link” was used to solicit stolen code.

  • Fraiture’s conduct predated his employment. OpenAI notes that Fraiture allegedly downloaded code while still at xAI and deleted it before joining OpenAI.

The answer also sets out fourteen affirmative defenses, ranging from failure to state a claim, waiver, and estoppel, to a notable fourteenth defense: that xAI’s claims are “frivolous, unreasonable, and brought in bad faith.” OpenAI specifically invokes 18 U.S.C. § 1836(b)(3)(D), which allows defendants in DTSA cases to recover attorneys’ fees where a trade secret claim is brought in bad faith.

Bad Faith Under the DTSA

One of the most strategically significant aspects of OpenAI’s filing is its reliance on the DTSA’s bad faith fee-shifting provision, § 1836(b)(3)(D), which authorizes courts to award reasonable attorneys’ fees to the prevailing party when a misappropriation claim is made “in bad faith,” when an injunction is sought or opposed in bad faith, or when misappropriation is willful and malicious.

What Does “Bad Faith” Mean Under the DTSA? 

The definition of “bad faith” under the DTSA is unsettled, and different circuits apply slightly different standards. In the Southern District of New York, Judge Liman’s recent opinion in Recoop LLC v. Outliers Inc., No. 22-cv-4535, 2025 WL 1725024 (S.D.N.Y. June 20, 2025), provides a very comprehensive overview. There, the court reviewed competing formulations across jurisdictions:

  • The Second Circuit has not adopted a uniform rule but generally follows the two-part meritlessness and improper purpose test, tracking the common law fee-shifting standard. See Insurent Agency Corp. v. Hanover Ins. Co., 2020 WL 86813 (S.D.N.Y. Jan. 9, 2020).

  • The Third Circuit requires proof that the plaintiff completely lacked evidence and knew, or was reckless in not knowing, that its claims lacked merit. See Elmagin Cap., LLC v. Chen, No. 22-2739, 2024 WL 2845535, at *5 n. 14 (3d Cir. Mar. 21, 2024).

  • The Seventh Circuit takes a broader approach, finding bad faith where claims are frivolous or maintained for improper purposes such as harassment or delay, or needless increase in the cost of litigation. See LQD Bus. Fin., LLC v. AKF, Inc., 2025 WL 830444, at *4 (7th Cir. Mar. 17, 2025).

  • The Sixth Circuit demands evidence that “that a party’s claim was meritless, that the party knew at a certain point that it was meritless and nonetheless maintained it, and that the party brought or maintained the claim for some improper purpose.” Shepard & Assocs., Inc. v. Lokring Tech., LLC, 2025 WL 1420931, at *4 (6th Cir. May 16, 2025).

  • The Fourth Circuit, by contrast, holds that a DTSA claim is not made in bad faith if it had “at least some chance of success,” and that a finding of bad faith “requires, at a minimum, that the plaintiff’s ‘claim had no chance of success under existing law.’ ” Akira Techs., Inc. v. Conceptant, Inc., 773 F. App’x 122, 125 (4th Cir. 2019) (quoting Tullidge v. Bd. of Supervisors of Augusta Cty., 391 S.E.2d 288, 290 (Va. 1990)).

In Recoop, Judge Liman concluded that most courts require both (1) objective speciousness (the absence of a reasonable factual or legal basis) and (2) subjective bad faith (an improper motive). Applying that framework, the court denied the defendant’s motion for fees, finding that the plaintiff’s weak but colorable claims did not satisfy that dual showing (“a failure of proof does not alone establish that the claim was brought in bad faith.”).

Illustrative DTSA Cases

The cases below further illustrate how courts apply (and often reject) bad faith fee requests:

  • TransPerfect Global, Inc. v. Lionbridge Technologies, Inc., No. 19-cv-3283 (DLC), 2022 WL 2119344 (S.D.N.Y. May 31, 2022): After granting summary judgment for defendants, the court denied fees, holding that although TransPerfect’s claims lacked evidentiary support, defendants failed to show the “high degree of bad faith” required under the DTSA.

  • Design Gaps, Inc. v. Hall, No. 3:23-cv-186-MOC, 2024 WL 203244 (W.D.N.C. Jan. 18, 2024): The court dismissed the DTSA claim for failure to identify any cognizable trade secret but declined to award fees, emphasizing that “transparently flawed” pleadings alone do not prove dishonest intent.

  • RJB Wholesale, Inc. v. Castleberry, 788 F. App’x 565 (9th Cir. 2019): The Ninth Circuit reversed a fee award, holding that even meritless trade secret claims do not justify fee shifting without evidence of improper motive or intentional abuse of process.

OpenAI’s Litigation Strategy

Bad faith fee awards under the DTSA remain rare and demand a high evidentiary showing. But OpenAI’s invocation of this defense is telling. It signals that OpenAI is preparing not only to defend the lawsuit but also to argue that xAI filed it to inflict reputational harm and deter employee movement. If successful, OpenAI could recover attorneys’ fees and set an influential precedent discouraging companies from using trade secret litigation as a competitive weapon in Silicon Valley’s AI arms race.

Why It Matters

If OpenAI succeeds in persuading the court that xAI’s claims were brought in bad faith, it could do more than win the lawsuit against it. A favorable ruling under the DTSA’s bad faith provision would help define how courts interpret and apply the DTSA’s fee-shifting provision and, in doing so, bolster the strategy for defendants to go on offense in trade secret litigation. Rather than simply defending against misappropriation allegations, OpenAI is using the DTSA to challenge the motives and methods behind the claim.

All eyes will now turn to the November 18 hearing before Judge Rita Lin, and we will keep this page updated accordingly.

Sarah Tishler is the author of this article. Sarah is a partner at Beck Reed Riden LLP. She is a commercial litigator whose practice focuses on complex business disputes in state and federal courts, including intellectual property, breach of contract, and fraud claims. Her experience also includes regulatory investigations, international arbitration, securities litigation, shareholder litigation, class-action litigation, and pre-transactional advising.

 

eck Reed Riden LLP is among the leading authorities in trade secret, noncompete, and unfair competition law, and our experience handling these matters is backed by our extensive employment law and business litigation experience. Our hand-picked team combines attorneys with complementary expertise and practical experience.

Beck Reed Riden LLP is Boston’s innovative litigation boutique. Our lawyers have years of experience working with clients ranging from Fortune 500 companies to start-ups and individuals. We focus on business litigation and employment.

We are experienced litigators and counselors, helping our clients as business partners to resolve issues and develop strategies that best meet our clients’ legal and business needs – before, during, and after litigation. We’re ready to roll up our sleeves and help you. Read more about us, the types of matters we handle, and what we can do for you here.

xAI v. Li Trade Secret Litigation Heats Up: Criminal Exposure and Fifth Amendment Battles

October 6, 2025
, AI, News, Trade Secret

The litigation storm surrounding xAI is no longer just a civil employment dispute. In the case against its former engineer, Xuechen Li, there are now high-stakes battles implicating both the civil trade secret misappropriation claim, and the federal criminal investigation that is also taking place. 

The Fifth Amendment and the Li Case

xAI’s case against its former engineer, Xuechen Li, took a dramatic turn when it emerged that federal prosecutors were investigating him for trade secret theft. According to Li’s filings, the day before xAI filed its civil complaint, the FBI executed coordinated search warrants on his residence, his vehicle, his person, and his hotel room. Agents seized three cell phones, multiple computers, notepads, notebooks, books, bank receipts, keys, and a flash drive. Li has been formally informed that he is the subject of a federal criminal investigation.

This backdrop has shaped the discovery fight. xAI obtained a temporary restraining order on September 2 requiring Li to turn over devices, accounts, and identify all storage locations where xAI’s confidential information might be found. But Li’s lawyers argue that compliance would effectively force him to admit the possession and movement of stolen data—admissions that could be used against him in a criminal case. In their briefs, they describe xAI’s requests as “wide-ranging, broad-sweeping discovery specifically designed to provide a ‘link in the chain’ to be used by law enforcement.”

Li has therefore invoked the Fifth Amendment against many of xAI’s interrogatories and production requests, particularly those requiring him to identify devices, accounts, or people to whom xAI data may have been disclosed. He argues that responding would not just be embarrassing or inconvenient but would directly map onto the factual questions prosecutors are pursuing.

Judge Rita Lin’s September 9 order struck a middle ground: Li must appear for deposition and respond to discovery, but he may assert the Fifth Amendment on a question-by-question basis. The court also authorized xAI to begin third-party discovery “without further delay.” The preliminary injunction hearing is scheduled for December 2, 2025.

The practical reality is stark: Li faces a choice between protecting himself in the criminal action with his silence, and the civil consequences of doing so. Refuse to answer, and xAI could argue adverse inferences; answer, and the responses could feed the criminal case. For xAI, leveraging that tension is now surely part of its litigation strategy, not just against Li, but against his new employer, OpenAI.

Why the FBI Is Involved: Theft of Trade Secrets Under 18 U.S.C. § 1832

To understand the gravity of the situation, it is instructive to look at how aggressively the U.S. government prosecutes trade secret theft. These aren’t run-of-the-mill civil cases — in many instances, they are high-stakes criminal prosecutions.

  • In May 2022, Dr. Xiaorong You (aka Shannon You) was sentenced to 168 months (14 years) in prison plus three years of supervised release and a $200,000 fine after a jury convicted her of conspiracy to commit trade secret theft, economic espionage, and wire fraud. Her offenses involved stealing proprietary formulations for BPA-free coatings during her work at major chemical companies.

  • On December 16, 2024, Klaus Pflugbeil was sentenced to 24 months in prison after pleading guilty to conspiring to send trade secrets belonging to a U.S.–based electric vehicle company.

  • In January 2022, Xiang Haitao, a Chinese national, pleaded guilty to conspiracy to commit economic espionage. His scheme involved the theft of proprietary software developed by Monsanto’s subsidiary to improve crop yields, with the intent of benefiting a foreign entity.

These prosecutions underscore how seriously federal authorities treat trade secret failures. Under § 1832, to obtain a conviction the government must prove that:

  1. The defendant improperly obtained or stole the information. 
  2. The defendant knew it was proprietary. 
  3. The information qualified as a trade secret. 
  4. The defendant intended to use it for someone other than the owner. 
  5. The defendant knew the owner would be harmed. 
  6. The trade secret related to a product or service in interstate or international commerce. 

Under § 1832, the sentence has the potential to be steep. Each count under § 1832 carries a maximum of 10 years in prison, along with substantial fines. That context makes it clearer why Li is fighting so hard to avoid admitting where he stored data, which devices he used, or whether information was transmitted internationally. Doing so could provide prosecutors the missing link in a criminal case against him.

The Broader xAI v. OpenAI Case

The developments in the Li case also implicate xAI’s broader lawsuit against OpenAI, which I wrote about in a previous post. That complaint alleges that OpenAI recruited Li and other employees to bring over Grok’s source code, inference systems, business strategies, and confidential data center deployment methods. If the civil case establishes that OpenAI knowingly benefited from misappropriated information, remedies could even include injunctions that affect OpenAI’s flagship products.

The Fifth Amendment issues in xAI v. Li may reverberate in the OpenAI case. If key witnesses refuse to answer discovery on criminal grounds, xAI will need to rely heavily on forensic evidence, third-party records, and circumstantial proof to connect the dots back to OpenAI. And if prosecutors bring a criminal case against Li, the effects could be dramatic: a conviction or guilty plea would essentially establish that xAI’s trade secrets were stolen, easing xAI’s burden in the civil action. It could also create reputational and regulatory pressure on OpenAI, even if the company is not a criminal defendant, by reinforcing xAI’s narrative of systematic misappropriation. Finally, evidence generated in a criminal prosecution (such as transcripts, forensic reports, or plea admissions) could provide xAI with powerful new material to use in its case against OpenAI.

In short, a criminal prosecution against Li could transform the OpenAI case. Today, xAI has to overcome Li’s Fifth Amendment silence with circumstantial evidence. If prosecutors secure a conviction or guilty plea, xAI would suddenly have a government-backed record that its trade secrets were stolen. That shifts the battleground in xAI v. OpenAI from “was there a theft?” to “what did OpenAI know, and when did it know it?”

Why It Matters

The FBI investigation into Li highlights the grim reality that in any high-stakes AI disputes, like high-stakes trade secret disputes more generally, the exposure may not just be civil liability; individuals may face criminal exposure. For xAI, the Fifth Amendment fights may delay discovery but also amplify the gravity of the underlying claims. For Li, silence may be the safest legal posture, but it comes with significant risk in the civil litigation.

And for OpenAI, the outcome of a potential prosecution against Li could be pivotal. A conviction would validate xAI’s claims of theft and shift the focus of the civil litigation away from whether xAI’s trade secrets were taken, and onto what OpenAI did with them. That reframing would dramatically strengthen xAI’s position, leaving OpenAI to defend not the existence of a theft, but its own knowledge and conduct.

As xAI presses forward against both Li and OpenAI, the fight over AI’s next frontier is playing out not just in labs and markets, but in courtrooms where the stakes include market dominance, criminal exposure, and the integrity of some of the most valuable trade secrets in the world.

 

Sarah Tishler is the author of this article. Sarah is a commercial litigator whose practice focuses on complex business disputes in state and federal courts, including intellectual property, breach of contract, and fraud claims. Her experience also includes regulatory investigations, international arbitration, securities litigation, shareholder litigation, class-action litigation, and pre-transactional advising.

 

eck Reed Riden LLP is among the leading authorities in trade secret, noncompete, and unfair competition law, and our experience handling these matters is backed by our extensive employment law and business litigation experience. Our hand-picked team combines attorneys with complementary expertise and practical experience.

Beck Reed Riden LLP is Boston’s innovative litigation boutique. Our lawyers have years of experience working with clients ranging from Fortune 500 companies to start-ups and individuals. We focus on business litigation and employment.

We are experienced litigators and counselors, helping our clients as business partners to resolve issues and develop strategies that best meet our clients’ legal and business needs – before, during, and after litigation. We’re ready to roll up our sleeves and help you. Read more about us, the types of matters we handle, and what we can do for you here.

xAI Sues OpenAI: Expanding the Trade Secret Battle in the AI Race

September 26, 2025
AI, News, Trade Secret

Just weeks after filing suit against a former engineer, xAI has escalated the dispute by suing OpenAI directly. In a complaint filed on September 24, 2025 in the Northern District of California, xAI accuses OpenAI of orchestrating a deliberate campaign to misappropriate xAI’s trade secrets, interfere with its economic relationships, and gain an unfair edge in the increasingly high-stakes race to dominate generative AI. You can read the full complaint here.

The Allegations

According to xAI, OpenAI deliberately targeted and recruited key employees — including engineers Xuechen Li and Jimmy Fraiture, and a senior finance executive — to obtain:

  • The complete Grok source code, including training and fine-tuning methods.
  • Infrastructure and inference systems unique to xAI’s production reliability.
  • Experimental code folders and future R&D directions from xAI’s founders and staff.
  • Business strategy and roadmap information, including an all-hands video recording of Elon Musk discussing contracts and priorities.
  • Data center deployment know-how, described internally as xAI’s “secret sauce,” and related supplier contract terms.

xAI claims OpenAI induced these employees to copy or retain such information by offering multimillion-dollar packages and communicating over encrypted channels like Signal to avoid detection.

Legal Framework and Case Law Context

The complaint asserts misappropriation under the Defend Trade Secrets Act (DTSA), alongside interference and unfair competition claims. How the court handles the trade secret allegations will hinge on two recurring doctrinal issues: identification and reasonable measures.

Identification: OpenAI may argue that xAI has not described its alleged trade secrets with sufficient particularity. However, the Ninth Circuit’s recent decision in Quintara Biosciences, Inc. v. Ruifeng Biztech, Inc., 2025 WL 2315671 (9th Cir. Aug. 12, 2025), makes clear that whether trade secrets have been identified with “sufficient particularity” is usually a fact question for summary judgment or trial. In reversing a district court’s dismissal, the Ninth Circuit emphasized that a DTSA trade secret claim “will rarely be dismissible” at the pleading or early discovery stage and that striking trade secrets prematurely was an abuse of discretion. That precedent may give xAI more leeway at this stage to proceed to discovery.

Reasonable measures: Courts require plaintiffs to show they took adequate steps to protect their trade secrets. In FrontRunner HC, Inc. v. Waveland RCM, LLC, No. CV 20-10230-DJC, 2020 WL 7321161, at *11 (D. Mass. Dec. 11, 2020), the court credited measures such as requiring NDAs as evidence of reasonable measures. But other courts have cautioned that confidentiality agreements alone are not enough. For example, in Diamond Power Int’l, Inc. v. Davidson, 540 F. Supp. 2d 1322, 1334 (N.D. Ga. 2007), the court held that requiring all employees to sign generalized confidentiality agreements “is generally not, standing alone, sufficient.” This creates a critical question: were xAI’s termination certifications, NDAs, and technical safeguards enough, or will the court find gaps that undercut its claims?

Why It Matters

By naming OpenAI itself, xAI is not just accusing former employees of disloyalty but alleging that one of the most prominent (if not the most prominent) AI labs is running a coordinated scheme to neutralize a competitor. If the allegations are substantiated, remedies could include injunctions requiring OpenAI to delete models trained with xAI’s stolen code, which would be a potentially industry-shaping outcome.

This case illustrates how trade secret law is becoming the primary battleground for AI competition. At one end, Quintara suggests courts will be reluctant to dismiss claims too early. At the other, cases like Diamond Power underscore that companies must pair agreements with robust technical and procedural safeguards. For AI developers, the lesson is clear: trade secret protection depends on both strong internal practices and the ability to prove them in court.

 

Sarah Tishler is the author of this article. Sarah is a commercial litigator whose practice focuses on complex business disputes in state and federal courts, including intellectual property, breach of contract, and fraud claims. Her experience also includes regulatory investigations, international arbitration, securities litigation, shareholder litigation, class-action litigation, and pre-transactional advising.

 

eck Reed Riden LLP is among the leading authorities in trade secret, noncompete, and unfair competition law, and our experience handling these matters is backed by our extensive employment law and business litigation experience. Our hand-picked team combines attorneys with complementary expertise and practical experience.

Beck Reed Riden LLP is Boston’s innovative litigation boutique. Our lawyers have years of experience working with clients ranging from Fortune 500 companies to start-ups and individuals. We focus on business litigation and employment.

We are experienced litigators and counselors, helping our clients as business partners to resolve issues and develop strategies that best meet our clients’ legal and business needs – before, during, and after litigation. We’re ready to roll up our sleeves and help you. Read more about us, the types of matters we handle, and what we can do for you here.

1 2 3 … 12