A trade secret you cannot prove existed, in a defined form, on a specific date, is not an enforceable trade secret. In any dispute, the burden falls on the holder to demonstrate what the secret was and when it existed — and internal records a company controls are rarely enough. In June 2026, a US court showed exactly how that gap collapses a case.
A business insures its premises, its equipment, its vehicles, its people. But the single most valuable thing many companies own is not on any of those policies: the confidential information that makes the business work — the source code, the technical specifications, the design files, the methods, the client data. It is often worth more than everything else combined, because it is the thing competitors cannot replicate. And it is routinely protected by nothing more than an assumption that internal records will speak for themselves if it ever comes to a fight.
On 15 June 2026, US District Judge Rita Lin permanently dismissed xAI's trade secret lawsuit against OpenAI in the federal court in San Francisco, ruling that a third attempt to plead the case would be futile. For most readers the story was another Musk-versus-OpenAI headline. For anyone responsible for protecting valuable information inside a business, it is a working illustration of how trade secret claims fail — and of the one thing every successful claim requires.
That one thing is prior proof of ownership: independent, verifiable evidence that a specific piece of information existed, in a defined form, on a specific date, and was treated as confidential. It is the foundation of every enforceable IP claim, and almost no organisation establishes it systematically.
This post explains why the xAI case collapsed, why the same evidence problem sits underneath almost every trade secret dispute in every major jurisdiction, what the law requires a holder to demonstrate, and what closing the gap requires.
The case failed on inducement, not on the secrets themselves. Judge Lin did not rule on whether OpenAI held any of Grok's secrets. She ruled that xAI had not pleaded facts showing OpenAI induced former senior engineer Xuechen Li to disclose those secrets. Asking a job candidate about their previous work, she reasoned, does not by itself amount to inducing the disclosure of a trade secret — to hold otherwise would expose employers to liability every time they interviewed someone with relevant experience.
The case had already been dismissed once, in February 2026; after xAI narrowed its allegations to a presentation Li gave during recruitment, the court dismissed it again on 15 June 2026, this time for good — three attempts across nine months, all defeated before the question of the secrets themselves was ever reached. That is one evidence problem: proving inducement. But underneath it sits an older and more universal one, and it is the one that should concern every business.
Before misappropriation is even argued, a plaintiff must prove the secret existed — and that is where most claims fail. Before any plaintiff can argue that a defendant misappropriated a trade secret, they must first prove three things: that the secret existed in defined, identifiable form at a specific time; that reasonable steps were taken to keep it confidential; and that the information derived independent economic value from being secret.
Most collapsed trade secret claims fail on the first of those, not because the secret did not exist, but because the plaintiff cannot prove, in evidence a court will accept, what specific information existed and when. Internal documents can be backdated. Email metadata can be manipulated. File-creation dates on a corporate server can be edited by anyone with administrative access. Employee testimony is, by definition, contested in a dispute. A court does not take a plaintiff's word that a document existed on a given date simply because the document says so — it requires independent corroboration, and the older the dispute, the harder that becomes.
What a business holds versus what a court needs
What a business typically holds | Why it is insufficient in a dispute |
Internal documents with dates | Can be backdated; created on systems the business controls |
Email metadata | Can be manipulated; not independently verified |
File-creation dates on a server | Editable by anyone with administrative access |
Employee testimony | Contested by definition once a dispute begins |
Every item on that list has the same weakness: the business itself controls it, so a court will not take its timing at face value. Prior proof of ownership removes that weakness. It fixes an immutable record of the exact moment a specific piece of information existed — created independently of the business, timestamped by a third party, and impossible to alter afterwards. That independently fixed moment of creation is the element most trade secret claims lack. In patent law it distinguishes invention from coincidence. In copyright it separates an original creator from an opportunist. In trade secret litigation it turns an allegation into an enforceable claim.
The requirement to prove a trade secret existed in defined form is not a quirk of one US courtroom — the same three-part burden is written into the law of every major jurisdiction. In the United States, the Defend Trade Secrets Act of 2016 created a federal civil cause of action and defines a trade secret as information whose owner has taken reasonable measures to keep it secret, and which derives independent economic value from not being generally known. A plaintiff who cannot identify the information with particularity, and fix it in time, fails at the threshold.
The European Union imposes the same structure. Article 2(1) of the Trade Secrets Directive defines a trade secret as information which is secret, has commercial value because it is secret, and has been subject to reasonable steps by the person lawfully in control of it to keep it secret. The United Kingdom carried the same definition into domestic law through the Trade Secrets (Enforcement, etc.) Regulations 2018, which operate alongside the older common law of confidence. In each regime the burden sits with the holder — and in each, the first question is identical: what exactly was the secret, and when did it exist in that form?
The consequence is that the evidential gap the xAI case exposed travels across borders intact. A UK business enforcing against a departing employee, a German manufacturer pursuing a supplier, and a California AI lab suing a competitor all face the same threshold demand for particularised, dateable proof. A protection strategy that cannot answer that demand fails everywhere at once.
Proving the secret existed is only the first limb — the holder must also prove they took reasonable steps to keep it secret, and courts examine those steps in detail. Tribunals on both sides of the Atlantic have weighed the same factors: whether access to the information was restricted to those who needed it; whether employees and partners signed confidentiality obligations that identified the material as protected; whether documents were marked confidential; whether technical controls such as access credentials and encryption were in place; and whether the business could show a consistent practice rather than a policy written after the dispute began.
The two limbs reinforce each other, and they fail together. A business that cannot produce a dated record of what the secret was will also struggle to show that its confidentiality measures covered that specific information at the relevant time. An NDA signed in 2024 protects the information that existed and was identified in 2024 — but if the business cannot prove what that information was, the NDA protects an abstraction. Reasonable steps taken over undefined material are steps a court cannot assess. The evidential record of what existed, and when, is the spine supporting every other protective measure.
Cryptographic timestamping supplies prior proof of ownership in a form the legal system can accept, and the mathematics is unambiguous. A cryptographic deposit generates an independent, verifiable record of exactly what a file contained and the moment it existed — an immutable timestamp secured across independent systems simultaneously, without the file itself ever leaving the holder's control.
How a cryptographic deposit works — in order:
Set that against the xAI fact pattern. A business holding such certificates for its confidential information does not have to rely on internal records a court will discount. It holds independent, verifiable evidence of exactly what existed and when — the precise element whose absence sinks most trade secret claims before the argument over misappropriation is even reached.
The problem is growing, not receding. Senior technical talent now moves between competing laboratories at a pace no previous generation of employers faced, and each departure carries knowledge that sits exactly on the boundary the xAI court examined: the line between general expertise, which an employee owns, and particularised secrets, which the business does. Every hire from a competitor is a potential future dispute in which both sides will need to prove what existed on their systems, in what form, before the move. The businesses that can answer with independent evidence will settle those disputes early and on favourable terms. The businesses that cannot will discover, as xAI did, that a claim without proof of the underlying secret never reaches the question of misappropriation.
Prior proof of ownership is the foundation that every other protection assumes already exists. A business will pay for specialist IP insurance — typically around £1,000 to £1,200 a year for a UK SME — and even that funds the legal fight on the assumption that the policyholder can already prove what they own. That assumption rests on prior proof.
Cryptographic prior proof is not insurance and does not pay out. It is the evidence a case depends on — the proof an insurer, a court, or a counterparty assumes is already in place. It is the difference between a trade secret that can be enforced and one that, in the eyes of a court, cannot be proven to have existed at all.
A record established today fixes the timeline today, with mathematical certainty. Constructed after the fact, in the middle of litigation, the same record is worth little — because its timing is exactly what the other side will dispute. The xAI case never reached the question of what the secrets contained. Most collapsed claims never do. The businesses that survive that first question are the ones that answered it before anyone asked.
NDA Breaches and the Evidence Problem: Proving What Was Disclosed Under an NDA
CPR 32.19: What UK Courts Require to Prove a Document Is Authentic
James Snell is the founder of Provlyn, a platform providing cryptographic prior proof of IP ownership. provlyn.com