Apple sued OpenAI in July 2026, alleging two former senior employees stole hardware trade secrets to accelerate OpenAI's push into consumer devices. In August 2026, Apple asked a federal judge for a preliminary injunction to block OpenAI and the ex-employees from touching that information; OpenAI responded by publishing internal iMessage screenshots and emails rebutting Apple's account. The case is now the most consequential trade-secrets fight of the AI era — and its outcome will shape the line between competitive hiring and IP theft for every AI company, every engineer, and every business that hires AI talent.
Last verified: 2026-08-05 • Apple filed the suit on July 10, 2026 (CNBC) and a preliminary injunction motion on August 4, 2026 (Reuters). • OpenAI's public rebuttal, including iMessage evidence, was posted August 3, 2026 (OpenAI company blog). • What it means for you: the case will redraw the rules on what AI engineers can take from one job to the next — especially across AI-hardware projects — and will affect timing of OpenAI's consumer device roadmap.
What is the Apple vs OpenAI trade secrets lawsuit about?
Apple alleges that OpenAI orchestrated a system to poach Apple hardware engineers and have them carry confidential iPhone, Apple Watch, and supply-chain know-how to OpenAI's consumer device division. The complaint names Tang Tan (Apple's former VP of product design for iPhone and Apple Watch, who later became OpenAI's Chief Hardware Officer) and Chang Liu (a former senior electrical engineer) as key defendants. Apple claims the pair downloaded confidential files, shared unreleased product information during hiring interviews, and that OpenAI instructed new hires on how to evade Apple's security protocols on departure (Wired; CNBC; TechSpot).
The suit was filed on July 10, 2026 in the U.S. District Court for the Northern District of California. Apple seeks damages, a permanent injunction, and an order requiring OpenAI to return all confidential materials. Less than a month later, on August 4, 2026, Apple escalated by filing a motion for a preliminary injunction — an emergency order that would freeze OpenAI and the individuals' use of the alleged secrets while the case proceeds (Reuters; 9to5Mac).
According to court documents and Wintel analysts' analysis of them, Apple frames ~400 former Apple employees as having made the jump to OpenAI, with many allegedly accessing confidential information tied to unreleased products before they left (TechSpot).
Who is Tang Tan and why is his role central?
Tang Tan is the ex-Apple vice president who led product design for the iPhone and Apple Watch over a 24-year career, and is now OpenAI's Chief Hardware Officer. He's at the center of Apple's complaint because he didn't just leave Apple — he co-founded io Products, the hardware startup OpenAI bought for nearly $6.5 billion in May 2025, and now runs the hardware division that competes directly with Apple's device roadmap (AppleInsider profile; The Verge). Tan spent 24 years at Apple, joined io Products in 2024, and was integrated into OpenAI when io and its ~55 employees were absorbed into OpenAI in the May 2025 acquisition.
What makes Tan an outsized legal target: Apple's complaint accuses him of (a) emailing himself supplier data during his final months at Apple, (b) using Apple confidential project code names during OpenAI hiring interviews, and (c) instructing new hires to bring unreleased Apple hardware components to job interviews to share (Wired; Amazonian analysis of public complaint). Tan has not been charged criminally. Civil trade-secret cases can be settled without the loser admitting liability.
How did OpenAI respond to Apple's injunction request?
On August 3, 2026, OpenAI published a company blog post titled "Apple is getting this wrong" calling Apple's filing "careless, aggressive and oddly personal" and asserting it "does not have, nor want, any of their trade secrets." The post laid out four rebuttals with exhibit-grade primary evidence:
- Misdelivered email. Apple's external counsel (Gabriel Gross of Weil, Gotshal & Manges) allegedly mixed up two similar Asian last names and emailed the wrong OpenAI contact to open the dispute. OpenAI's GC Che Chang flagged the caller's misrepresentation ("this guy from Weil is lying about speaking with me on the phone"), and Apple heard nothing more for five months.
- Invented conversation. Apple claimed it had a phone discussion with OpenAI's General Counsel that OpenAI says never happened. Apple later conceded the call did not occur.
- Post-departure contact evidence. OpenAI published iMessage screenshots of Apple employees asking Chang Liu for help finding files and decoding Apple infrastructure after his last day (January 22, 2026). On March 5, 2026, an Apple employee messaged Liu asking him to walk through how an Apple subsystem powered downstream items — at which point another employee interjected, "Hi, this is highly irregular, please remove me from this thread."
- OpenAI's policy on competitor IP. OpenAI states Tang Tan was explicitly instructed that OpenAI does not want and must not use confidential information from other companies, against the corporate wall it operates.
We do not have, nor want, any of their trade secrets. We're much more interested in building innovative products and technologies that push the frontier. — OpenAI, August 3, 2026
What about Chang Liu — what is "residual access"?
Apple's complaint accuses Chang Liu of retaining access to Apple files and exploiting what it called a bug in Apple's authentication system after his departure. OpenAI rebuts with two points: (a) that Apple itself first reached out to Liu after he left, asking for help locating files and explaining its own internal systems, and (b) that what Apple calls theft is actually "residual access" — Apple's well-documented failure to revoke file permissions when employees leave. From OpenAI's account, ex-Apple workers routinely retained Apple file access without knowing it until they were explicitly told, and at least one such incident here went up the chain only in 2026.
Apple has not yet addressed the "residual access" framing publicly — when Apple files a reply the court will hear its version of the same story. Either way, the dispute exposes the operational risk that any company with deferred-offboarding faces: failure to revoke credentials at separation is now a litigation liability. For an AI company building a hardware division and hiring hundreds of workers from a competitor, the decision to "offer oportunenya swoop mode" — even bogusly managing-the-exit — is high-risk behavior.
What is a preliminary injunction and why does Apple want one?
A preliminary injunction in a trade-secrets case is a court order issued before the trial ordering a defendant to avoid using or disclosing the alleged secrets while the case is pending. Apple argues it "will be irreparably harmed absent a preliminary injunction" — the standard phrase invoking the four-factor federal test: (1) likelihood of success on the merits, (2) irreparable harm without the order, (3) balance of harms favoring the applicant, and (4) public interest.
The Defend Trade Secrets Act of 2016 (DTSA) created the federal civil cause of action Apple relies on. Under the DTSA, mere acquisition of trade secrets by improper means — before any use — is enough for liability. The Supreme Court's long-running Winter v. Natural Resources Defense Council standard is what governs each factor above. For tech companies, the practical question is never "did OpenAI win the case"; it's that even a narrow injunction freezing a few prototypes or certain supply-chain contacts can be catastrophic when a company is racing a hardware roadmap.
Apple is also asking for expedited discovery (depositions of Tang Tan, Chang Liu, named OpenAI staffers, plus corporate representatives from OpenAI and its wholly owned subsidiary io Products). Expert commentary suggests an injunction hearing will likely happen in late 2026 and that the evidence released in expedited discovery may push both parties toward a settlement (via DailyJournal, legal commentary by Prof. Elizabeth A. Rowe and Orly Lobel).
Why does the io Products acquisition matter to the case?
The io acquisition is the engine of Apple's case — it's the alleged mechanism by which trade secrets moved to OpenAI. io Products was co-founded in 2024 by Tang Tan and Jony Ive (Apple's legendary former design chief). In May 2025, OpenAI bought io in an all-stock transaction valued at close to $6.5 billion — then OpenAI's largest acquisition ever — and absorbed io's roughly 55 engineers, designers, and manufacturing experts into what became OpenAI's hardware division (The Verge; analyst round-up). Tan became OpenAI's Chief Hardware Officer; Ive's design firm LoveFrom took over design across OpenAI's products.
Apple's Shutdown-Myspace-settlement love letter specifically frames io as the vehicle that was used to "free-ride off Apple's decades of innovation" — the complaint alleges OpenAI "turned to trade-secret misappropriation to free-ride off Apple's hundreds of billions of R&D," citing supplier ties to Foxconn, Luxshare, and Goertek (which overlap heavily with Apple's own supplier network). io's first hardware prototypes were reportedly shown in late 2025; a HomePod-style home device with a camera and mechanical motion is rumored for a 2026 reveal with a 2027 ship target (complaint analysis, Bloomberg via AppleInsider).
What is "negative know-how" and why is it the riskiest part of the case?
The most legally novel part of Apple's complaint addresses "negative know-how" — knowing what doesn't work. Apple argues that even when ex-employees don't take specific documents, the avoidance of dead ends they learned at Apple accelerates OpenAI's development. This is the part of the case that, if Apple wins, will matter most to the AI industry: it turns knowledge acquired during years of iterative design into something a former employee may have to actively suppress when working for a competitor.
For a small business or AI startup, this is the takeaway: if you hire a senior engineer from a competitor, the line between "valuable experience" and "misappropriated trade secret" is not where you'd put it. University of San Diego law professor Orly Lobel put it: "A company may hire someone because that person understands hardware engineering, product development, suppliers, or manufacturing. The legal line is crossed when the new employer seeks the former employer's actual secret materials or asks questions calculated to elicit proprietary information" (DailyJournal). Stanford law professor Mark Lemley agreed: hiring many employees isn't illegal in California and may be good practice; but if employees actually took confidential documents and OpenAI is using them, "that is a problem," he said.
What is OpenAI being accused of, specifically — in plain English?
Apple's complaint alleges five categories of trade secrets were unlawfully taken:
| Category | What Apple alleges was leaked | Why it matters commercially |
|---|---|---|
| Unreleased product specs | Drawings, internal specs for unreleased iPhone / Watch / form-factor work | Lets OpenAI skip months of iteration |
| Manufacturing partnerships | Foxconn, Luxshare, Goertek supplier contacts and cost knowledge | Lets OpenAI sign faster, cheaper manufacturing deals |
| AI/ML hardware integration | How Apple integrates LLMs into hardware profiles | Directly overlaps OpenAI's device roadmap |
| "Negative know-how" | Knowledge of what doesn't work | Avoids expensive dead ends |
| Security/post-employment protocols | Apple's internal offboarding and access revocation design | Taught OpenAI which controls to "evade" |
Source: CNBC complaint summary; explainx.ai complaint analysis.
Note OpenAI has formally denied all allegations and argues the factual record in Apple's brief is selectively incomplete (no criminal charges have been filed; OpenAI has not been convicted of anything).
How does this compare to recent trade-secrets cases in tech?
| Case | Year | Outcome | Lesson |
|---|---|---|---|
| Waymo v. Uber | 2017 | $245M settlement; engineer Anthony Levandowski referred for criminal probe | Hidden lessons: acquisition-hiding + stolen LiDAR files moved to a competitor |
| Rivos v. Apple | 2022→2024 | Settled (terms undisclosed 2024) | Apple sued a startup founded by ex-employees over chip trade secrets |
| Apple v. OpenAI | 2026 | Pending; injunction hearing expected late 2026 | Public-PR-counterattack pattern (OpenAI published exhibits fast) |
| Tesla v. Zoox / Rivian | 2018–2024 | Tesla settled with Zoox 2024; Rivian case settled | Pattern of suits over ex-employees and Autopilot trade secrets |
Apple has been a serial plaintiff in trade-secrets cases — Rivos (2022), a 2022 case against ex-employee Simon Lancaster, and a 2019 case against Xpeng — and seems in 2026 to be using litigation as a "delay tactic against a hardware competitor it cannot outbuild fast enough" (explainx.ai analysis).
What this means for you: AI builders, businesses, and engineers
If you're an AI company hiring competitors' engineers
- Onboard with structural competitive-IP hygiene. Document the instruction you give new hires about not bringing confidential materials; collect written acknowledgements. OpenAI's published instruction to Tang Tan — "we don't want and must not use confidential information from other companies" — is the exact documented instruction that produces the paper trail litigators first ask for.
- Audit offboarding. If you lose an employee their old employer will look at your own offboarding practices too. Failure to revoke access is now a litigation liability, not just an ops one. Auditable explicit revocation of source-control, file share, and supplier-portal access is the cheapest legal-ion-shield you have.
- Track supplier overlap. If a new senior hire worked at the same Tier 1 supplier as a former employer, your supplier agreements — for those suppliers, on both sides — will be one of the most discovery-sensitive documents you have.
If you're an AI engineer moving between competitors
- Don't email yourself company files. Even for "reference later" or portability, this almost always flags case-defining episodes. Negotiate hard for the company's IP/exit policy in writing — before signing a competing offer.
- Refuse to answer interview questions about a former employer's non-public designs, plans, or firmware. Stanford's Lemley distinguishes between "experience" and "the employer seeking information covered by confidentiality and asking someone to breach a duty" — the latter is where civil and potentially criminal liability starts.
- Get your own counsel if asked to testify. Apple's motion asks for depositions from Tang Tan, Chang Liu, and possibly even an unnamed OpenAI staffer who once worked at Apple. Engineer statements — voluntary or compelled — are likely to be the precedent-setting parts of this case.
If you're a business betting on OpenAI's hardware roadmap
- The rumored device is a screenless, voice-first home AI companion with a camera, mechanical motion, and "GPT-Live" voice — not a smartphone killer (per Bloomberg via AppleInsider). Target unveil: 2026, ship: 2027.
- Any preliminary injunction that freezes specific prototype or supplier relationships can delay that timeline. Investors and customers should not plan around a firm ship date until the injunction hearing resolves.
- For the broader consumer-AI-hardware market, the case increases the regulatory and litigation cost of competitor-talent fly-overs — which likely favors large incumbents that can afford "soft legal" set-offs (Apple, Google, Microsoft) and slows pure-play startups trying to ship devices fast.
FAQ
Q: What did Apple sue OpenAI for? A: Apple sued OpenAI on July 10, 2026 in the Northern District of California, alleging trade-secret theft via former Apple employees Tang Tan and Chang Liu. Apple claims OpenAI built its consumer-hardware roadmap using misappropriated iPhone, Watch, and supply-chain secrets, and seeks damages plus an injunction.
Q: What is a preliminary injunction in a trade-secrets case? A: A preliminary injunction is a pre-trial court order freezing a defendant's use of the alleged trade secrets while the case proceeds. Apple filed this motion on August 4, 2026 arguing it would "suffer irreparable harm" without one. The federal test (from Winter v. NRDC) asks for likelihood of success, irreparable harm, balance of harms, and public interest.
Q: Is OpenAI guilty of stealing Apple's trade secrets? A: OpenAI has denied all allegations and published primary-source evidence (iMessages, emails) suggesting Apple's factual account is incomplete. As of August 5, 2026 there has been no trial, no verdict, and no criminal charges. Apple's complaint is a civil accusation; OpenAI's denial is the rebuttal; the court has not ruled on the merits yet.
Q: Who is Tang Tan and how is he involved? A: Tang Tan is Apple's former vice president of product design for the iPhone and Apple Watch (a 24-year career) and now OpenAI's Chief Hardware Officer after co-founding io Products, the startup OpenAI acquired for ~$6.5 billion in May 2025. Apple accuses him of emailing himself confidential supplier information and using Apple code names in OpenAI hiring interviews.
Q: What is the io Products acquisition and why is it central? A: io Products was the hardware startup co-founded by Tang Tan and Jony Ive in 2024. OpenAI bought it for ~$6.5 billion in all stock in May 2025, absorbing ~55 engineers into what became OpenAI's hardware division. Apple's complaint frames io as the vehicle for the alleged trade-secret transfer.
Q: How much could OpenAI owe if it loses? A: If Apple prevails, OpenAI could face a permanent injunction barring use of the trade secrets, monetary damages (potentially multi-million), and possibly exemplary double damages under the DTSA for willful misappropriation. The more immediate risk is a temporary injunction that freezes specific work streams and delays the consumer-device roadmap.
Q: Does this affect ChatGPT or OpenAI's current software products? A: No immediate operational effect on ChatGPT or the API. The lawsuit targets OpenAI's hardware division and the io Products subsidiary — not OpenAI's language models or API services. However, the broader message — that talent movement between big AI labs is now litigation-laden — does affect any team trying to lock-in founder-level AI engineers from competitors.
Discussion
0 comments