Apple's OpenAI Lawsuit Isn't a One-Off. It's a Pattern With a Paper Trail.

Last updated July 13, 2026. This is a developing legal story and will be updated as the case progresses.

Tang Tan spent years inside Apple's hardware organization before he left to build devices for OpenAI. According to Apple's new federal lawsuit, he never really stopped mining that job. The complaint alleges Tan used Apple's own internal codenames to draw more information out of job candidates during interviews, and that he told at least one candidate still employed at Apple to bring an actual battery, logic board, and other components to a "show and tell" session. Another former Apple engineer, the suit claims, kept a company laptop after he left, found a bug that let him back into Apple's cloud storage months later, and messaged a former colleague something close to delighted disbelief that it worked.

That's the sworn version of events now sitting in the U.S. District Court for the Northern District of California. It would be a serious story on its own. What makes it more than that is timing: a different federal court had already found, months before Apple filed a single page, that this same executive likely engaged in a strikingly similar pattern against a completely different company. Apple isn't the first party to accuse Tang Tan of this. It's the second, and OpenAI's hardware division is now facing overlapping trade secret allegations from two directions at once, on top of a three-year run of partners, collaborators, and former employees who each came away from OpenAI feeling like the trust ran one direction.

What Apple's Lawsuit Actually Alleges

Apple filed suit on July 10, accusing OpenAI, Tan, and former Apple electrical engineer Chang Liu of trade secret misappropriation and breach of contract. The complaint says the conduct reaches "at every level," from OpenAI's technical staff up to Tan, who now serves as the company's chief hardware officer. Apple also disclosed, for the first time publicly, that more than 400 former Apple employees now work at OpenAI.

The specific allegations go well beyond ordinary poaching:

• Tan allegedly used Apple's internal project codenames during OpenAI job interviews to draw more confidential detail out of candidates who still worked at Apple.
• Candidates were reportedly directed to bring "actual parts," including batteries, logic boards, and system-in-package components, to interview sessions for "show and tell."
• One candidate is accused of screenshotting and downloading files related to a confidential Apple project hours before an interview with Tan, who then allegedly asked follow-up questions about that same project once the interview began.
• Apple says Tan possessed and distributed an internal Apple document called a "Need to Know" list to new OpenAI hires before they had even given notice at Apple.
• Chang Liu, an eight-year Apple systems electrical engineer, allegedly kept a company-issued laptop after leaving for OpenAI, discovered a bug that let him access Apple's internal network storage, and downloaded confidential technical documents, later telling a still-employed Apple colleague he found the access "so funny."
• Apple alleges OpenAI used a proprietary Apple metal-finishing technique after misleading a manufacturing partner into believing Apple had authorized its use.

Apple says it first raised these concerns directly with OpenAI in February and received no response, which is part of why the filing calls the disclosed conduct "the tip of the iceberg." Apple is seeking damages, injunctive relief, and a court order barring OpenAI from using the disputed information. OpenAI has denied wrongdoing, telling reporters it has "no interest in other companies' trade secrets."

The lawsuit lands at an unusually awkward moment for the two companies. ChatGPT has been integrated into iOS since 2024, meaning Apple is currently suing a company whose product ships inside its own operating system.

Tang Tan Has Been Here Before: The iyO Case

Here's the detail that turns Apple's complaint from an isolated dispute into evidence of a pattern: Tang Tan was already the named defendant in a separate trade secret case before Apple ever filed.

iyO Inc., a small Bay Area startup building AI-powered earbuds, sued OpenAI, Sam Altman, Jony Ive, and Tan in 2025 over the name "io," which OpenAI used for the $6.5 billion hardware venture it acquired from Ive. The dispute started as a trademark fight. iyO argued the names "io" and "iyO" were phonetically identical and that OpenAI's high-profile launch, complete with a nine-minute Altman video calling the device "the coolest piece of technology the world will have ever seen," was actively confusing its investors and threatening its ability to raise money. A federal judge agreed, issuing a temporary restraining order in June 2025 that the Ninth Circuit later upheld on appeal.

Then, in March 2026, months before Apple's suit, iyO amended its complaint to add something more serious than a naming dispute: trade secret misappropriation and corporate espionage claims against Tan by name, identifying him specifically as Apple's former vice president of product design. In April 2026, a federal judge granted iyO a preliminary injunction, finding the company likely to succeed on its trademark claim and letting the trade secret allegations proceed. iyO's founder, Jason Rugolo, has publicly accused Tan and OpenAI of investigating his company and then "opportunistically" targeting it with a competing product launch built on information gathered under false pretenses.

Strip away the branding dispute and the shape underneath is close to identical to what Apple now alleges: an OpenAI hardware executive accused of using access, relationships, or interviews to extract confidential information from a smaller party, followed by a product built on the back of it. One company is a startup that had to fight in court just to keep its own name. The other is the largest technology company in the world. Both ended up making the same accusation against the same person within roughly a year of each other.

The Pattern Predates Apple: Johansson, Musk, and Microsoft

Widen the frame further and Apple's lawsuit stops looking like an outlier at all.

Scarlett Johansson. In 2024, OpenAI approached the actress twice asking to license her voice for ChatGPT's new voice assistant. She declined both times. Two days after her second refusal, OpenAI unveiled a voice called Sky that friends of Johansson's and news outlets reportedly could not distinguish from her own. OpenAI insisted Sky belonged to a different actress hired before any outreach to Johansson, and pulled the voice within a week once her attorneys got involved. A later independent analysis comparing Sky's vocal characteristics against more than 600 professional actors found it statistically closer to Johansson than to 98 percent of the sample. The person who said no ended up being the person OpenAI used anyway, with just enough procedural distance to argue about it publicly.

Elon Musk. OpenAI's founding donor sued the company over what he described as its abandonment of the nonprofit mission he helped fund in its earliest days. The case went to trial and a jury sided with OpenAI in May 2026, ruling that Musk waited too long to bring his claims. Musk has said he'll appeal. Whatever the legal outcome, the underlying grievance tracks the same shape as everything else on this list: someone who helped build OpenAI's early credibility watched the company's actual conduct drift from what he believed he'd signed up for.

Microsoft. When OpenAI's board fired Sam Altman in November 2023, Microsoft, OpenAI's largest financial backer, got almost no advance notice and watched its own stock drop on the news. Within 48 hours, Satya Nadella had hired Altman and co-founder Greg Brockman to run a new internal AI lab at Microsoft, only to lose both of them back to OpenAI two days later once employee revolt forced the old board to reverse course. Microsoft absorbed the whiplash of a five-day governance crisis at a company it had already poured billions into, with no seat at the table to prevent it. It came away with a non-voting board observer role, which says something about how much influence its money actually bought.

That imbalance never fully resolved. Through 2025 and into 2026, OpenAI's compute needs outgrew what Microsoft alone could supply, and OpenAI struck deals with AWS, Oracle, and SoftBank that put it in direct tension with the exclusivity Microsoft believed it had locked in. The two companies spent nearly a year in what multiple outlets described as contentious negotiations before restructuring their agreement in April 2026, stripping Microsoft's cloud exclusivity and narrowing its intellectual property rights, even though Microsoft holds a 27 percent stake in OpenAI's for-profit arm worth roughly $135 billion. The company with the single largest financial exposure to OpenAI still ended up renegotiating from behind, catching up to decisions OpenAI had already made elsewhere.

The Pattern Isn't Just External

It would be easy to read all of this as a story about how OpenAI treats outside partners. The 2024 non-disparagement scandal shows the same instinct pointed inward, at its own people.

Vox reported that spring that OpenAI had been requiring departing employees to sign lifelong non-disparagement agreements as a condition of keeping equity they had already vested, agreements so restrictive that even acknowledging their existence counted as a violation. Leaked documents showed the equity-cancellation language had been reviewed by senior leadership, including Altman himself, contradicting the company's initial framing of the practice as a paperwork oversight. The scandal broke the same week two of OpenAI's most senior safety researchers, Ilya Sutskever and Jan Leike, resigned, with Leike writing publicly that "safety culture and processes have taken a backseat to shiny products." Altman apologized, called himself "genuinely embarrassed," and said the company would strike the clawback language from its exit paperwork. He also said he hadn't known the provision existed, a claim the leaked documents partly undercut.

The through-line connecting a non-disparagement clause aimed at departing researchers, a voice actress who said no, a founding donor, a primary investor, a startup fighting for its own name, and now Apple's confidential hardware roadmaps is not a legal theory. It's a posture: agreements and boundaries function as opening positions at OpenAI, not as commitments that hold once they become inconvenient.

Why This One Is Different

Trade secret cases are notoriously hard to win outright, and companies poach talent from each other constantly without ending up in federal court. Apple's defenders inside the industry will point that out, correctly. But three things separate this case from the pattern that preceded it.

First, it's Apple, not a startup or an individual, bringing the weight of the largest technology company in the world and a legal team with essentially unlimited resources. Second, it arrives with a nearly identical precedent already in progress: a federal judge has already found a startup likely to succeed on trade secret claims against the same executive OpenAI now has running its hardware business. Third, the timing lands squarely on top of OpenAI's own consumer hardware ambitions, the device Tan was hired to help build, which has already been delayed past its original branding, renamed internally to "Sweetpea," and pushed to a 2027 shipping window. A company trying to convince the market it can ship credible hardware is now defending, in parallel, two separate federal lawsuits alleging its hardware chief built that program on other companies' confidential work.

If even a portion of Apple's allegations hold up in discovery, "OpenAI has a habit of testing boundaries with people who trust it" stops being a fair characterization and starts looking like documented practice, repeated with enough specificity and enough named individuals that "OpenAI trade secret lawsuit" is no longer a single headline. It's a category.

What to Watch Next

This case is early. Discovery hasn't started, OpenAI hasn't filed a substantive response, and the iyO trade secret claims against Tan are still working through the same Northern District of California courthouse. A few things worth tracking as both cases develop: whether Apple's complaint produces internal OpenAI communications that corroborate the "Need to Know" document allegation, whether the iyO and Apple cases end up sharing discovery or witnesses given the overlapping claims against Tan, and whether OpenAI's hardware timeline slips further while its chief hardware officer is defending two federal suits at once. We'll keep this piece updated as the record develops.