apple just sued openai: inside the trade secret heist that ended the partnership era

Apple's lawsuit alleges a coordinated campaign to steal product roadmaps, and it names the engineer who did it. — Generated via ComfyUI / Z-Image Turbo

Apple filed suit against OpenAI on July 10, 2026, alleging a coordinated campaign to misappropriate confidential product information. The complaint, lodged in the Northern District of California, names former Apple designer Tang Tan and accuses him of using internal Apple code names in messages to OpenAI recruiters. The case alleges that OpenAI targeted at least six Apple employees, that the sought-after materials included product roadmaps, and that the effort was directed at accelerating OpenAI's hardware and consumer-device programs. TechCrunch's filing on the case describes it as a "trade secret heist." That phrasing is closer to the complaint than the press release. The rest of this article walks through the document.

The Filing: What Apple Actually Put in the Complaint

Apple's complaint is a 47-page document. It pleads four causes of action: misappropriation of trade secrets under the Defend Trade Secrets Act, breach of fiduciary duty, tortious interference with contract, and unfair competition under California Business and Professions Code §17200. The trade secret count is the spine. The other three are accessory.

The complaint alleges that OpenAI, beginning no later than early 2025, identified Apple employees with access to "confidential hardware and software roadmaps, including unreleased product specifications, internal design documents, and personnel and organizational information." It alleges that the targeting was systematic, that the recruiter on the OpenAI side was identifiable, and that the approach used "specific, non-public identifiers" — that is, internal code names — to establish credibility with the Apple employees.

The complaint also names Apple's own internal controls, but in a defensive posture. Apple says it has policies that prevent former employees from retaining confidential materials. The implication is that the controls failed, or that the materials were taken before exit, or both. The complaint does not specify which.

This is a specific kind of lawsuit. It is not a patent case. It is not a pure breach of contract case, although the breach counts are there. It is a case about people, and the inferences that can be drawn from their communications.

The Named Man: Tang Tan, the Recruiting Trail, and the Code Names

Tang Tan joined OpenAI in September 2025. At Apple, he was a senior director of product design. He had spent more than a decade at the company. His public LinkedIn describes work on "consumer hardware and accessories." His non-public work, according to the complaint, included product lines that Apple had not announced.

The code names are the most damaging part of the filing. Apple's complaint includes exhibits showing messages in which Tan allegedly referred to specific Apple projects by their internal code names — names that, by Apple's account, appeared nowhere outside Cupertino. The complaint does not print the code names in the public version of the filing. It does describe them as "unannounced hardware projects," "an internal AI initiative," and "a confidential operating-system effort."

Using a code name in a recruiting conversation accomplishes two things. It establishes that the speaker has the access they claim to have. It also tells the listener what the speaker knows. The combination is, in trade-secret litigation, almost always treated as a confession.

The complaint alleges that Tan was not the only target. It identifies at least five other Apple employees who received similar approaches, and it alleges that the OpenAI recruiter coordinating the campaign was a single individual whose name is redacted in the public filing. The inference is that the campaign was designed to be deniable at the level of any single message, but legible in aggregate.

The Alleged Haul: Which Apple Programs OpenAI Supposedly Wanted

The complaint describes three categories of confidential information that OpenAI is alleged to have sought.

First, product roadmaps. Apple alleges that the information sought included the timing, feature set, and technical specifications of unannounced hardware products. The complaint does not identify the products by name, but the description — "consumer hardware," "accessories," and a program with "an on-device AI focus" — points to a specific cluster of Apple's pipeline.

Second, internal AI initiatives. Apple has, since 2024, been publicly reorganizing around on-device AI. The complaint refers to a "confidential AI initiative" and to "internal model architecture documents." The implication is that OpenAI was interested in Apple's on-device model work, not in its chatbot integration.

Third, organizational information. The complaint alleges that OpenAI sought "personnel and organizational information," including details of Apple's AI team structure. This is unusual. It is more characteristic of a competitor building a poaching campaign than of a partner evaluating a technology integration.

The complaint does not allege that any trade secret was successfully exfiltrated to OpenAI's systems. It alleges that the information was communicated in messages. The legal question is whether the act of communication, without confirmed exfiltration, supports the trade-secret claim. Under the DTSA, it does.

The Partnership Paradox: Suing the Model You're Shipping in Siri

Apple and OpenAI announced an integration of ChatGPT into Siri in mid-2024. The integration shipped in late 2024 and has been updated since. The integration is a commercial relationship. Apple pays OpenAI for inference, and OpenAI gets distribution to Apple's installed base.

The lawsuit is filed against that same relationship. Apple is, in effect, suing the company whose model it ships to its users, while continuing to ship that model to its users.

The complaint does not address the integration. It does not seek to terminate the commercial relationship. It does not seek an injunction. It seeks damages and an order requiring OpenAI to destroy any confidential information in its possession.

This is consistent with a position that Apple is willing to live with. It is also consistent with a position that Apple is willing to negotiate from. The relationship is too valuable, in either reading, to break over a single lawsuit. The lawsuit is leverage.

The deeper implication is that the partnership is now conditional. Apple has, by filing, made clear that it believes OpenAI is engaged in a pattern of conduct rather than an isolated incident. The next integration, the next model update, the next contract renewal will all occur against the backdrop of a pending trade-secret action. That is not a stable position for either party.

What This Breaks: OpenAI as a Neutral Platform Is Now a Legal Target

OpenAI has, since 2023, presented itself as a neutral platform — a model provider that supplies infrastructure to other companies without competing with them at the application layer. The strategy has been commercially successful. It has also been legally valuable. Neutrality is a defense against certain antitrust theories, and a posture that reduces friction in enterprise sales.

The Apple complaint does not call OpenAI a competitor in name. It does not need to. The factual allegation is that OpenAI, in parallel with supplying Apple, was recruiting Apple's employees to work on overlapping programs, with knowledge of Apple's confidential plans. That is the conduct of a competitor. The complaint says so obliquely. The day's news cycle, summarized in the new-horizon.tech daily digest, included three separate reports of AI companies pressuring partners' employees. The pattern is broader than OpenAI. A separate NYT report on the same day documents parallel pressure on Meta from generative image providers.

For OpenAI, the cost is not the damages. The cost is the loss of posture. A platform that is also a competitor is a platform that is a target. The next enterprise customer that evaluates OpenAI will, after this filing, have a new set of questions. The next AI startup that considers OpenAI as a model provider will have a new set of hesitations. The legal exposure now extends to every conversation between an OpenAI recruiter and an employee of a partner.

The Discovery Cliff: Depositions, Chat Logs, and the First AI Trade-Secret Trial

Trade-secret cases turn on discovery. The pleadings are scaffolding. The substance is in the documents.

Apple will seek, and is likely to obtain, the communications of the named OpenAI recruiter, the communications of Tang Tan during the period in question, and the communications of the OpenAI executives who directed the recruiting effort. The scope will extend to Slack, iMessage, and Signal messages. It will extend to the internal notes tied to the candidates. It will extend, depending on the judge's rulings, to the deliberations of OpenAI's board about the consumer hardware program that Tan allegedly joined to support.

The legal question that will dominate the next twelve months is the scope of the protective order. OpenAI will argue that its internal communications about a consumer hardware program are competitively sensitive. Apple will argue that the communications about its own employees are not. The judge will decide.

If the case goes to trial, it will be the first major trade-secret action against a frontier AI company. The trial will not be about the technology. It will be about what the recruiters said, what the engineers said, and what the executives knew. The exhibits will be chat logs. The witnesses will be recruiters. The verdict will turn on the meaning of a code name used in a message at 11:47 p.m. on a Tuesday in March 2025.

That is the discovery cliff. The complaint is the visible edge. What is on the other side is not yet knowable.