Apple Sues OpenAI, and the Hardware War Turns Legal

The AI industry's talent war has been fought so far in the language of signing bonuses and LinkedIn announcements — a departure here, a poaching there, the occasional viral resignation. On July 10, Apple moved the fight into a federal courtroom. In a suit filed in the Northern District of California, Apple accused OpenAI of stealing its trade secrets, and it did not hedge the language: the theft, Apple's filing alleges, ran "at every level, from members of its Technical Staff to its Chief Hardware Officer." This is no longer a war for people. It's a war over what those people carried out the door.

The specific accusations

Apple's complaint is unusually granular for an opening filing, and the details are what make it dangerous. The company alleges that OpenAI's hardware chief Tang Tan — himself a former Apple vice president — directed Apple employees interviewing at OpenAI to hand over Apple secrets as part of the interview process, in at least one instance asking candidates to bring "actual parts" from Apple to their meetings. It alleges OpenAI coached departing Apple staff on how to evade the security procedures Apple runs when an employee leaves. And it names a former employee, Chang Liu, whom Apple accuses of walking out with a company laptop.

The material allegedly extracted wasn't abstract know-how. Apple describes technical specifications, engineering presentations, and proprietary project data covering unannounced technologies, features, and products. If even a fraction of that is substantiated, it moves the case out of the murky territory where trade-secret disputes usually live — where an employee's general skill and an employer's confidential information blur together — and into something closer to documented exfiltration. Asking a candidate to bring physical parts to a job interview is not a gray area. It's the kind of specific, deliberate act that discovery is built to surface.

OpenAI, for its part, has denied everything, stating it has "no interest in other companies' trade secrets." That is the only response available at this stage, and it will be tested against whatever internal communications and device forensics the litigation drags into the light.

Why hardware, and why now

The suit doesn't come out of nowhere — it comes out of a partnership going sour. Apple and OpenAI struck a high-profile deal in 2024, wiring ChatGPT into Apple's ecosystem in the arrangement that defined that year's AI-on-device narrative. Relations chilled for a reason any strategist could have predicted: OpenAI stopped being purely a model supplier and started becoming a hardware competitor. Its $6.4 billion acquisition of Jony Ive's io Products — a firm built by the designer most responsible for the iPhone's physical form — announced OpenAI's intent to build consumer devices of its own. Apple named io Products in the suit too, which tells you the company sees the hardware ambition, not the model, as the threat.

That framing matters. Apple is not fundamentally worried about OpenAI's language models; it partnered with them. It is worried about a company assembling, piece by piece, the exact capability Apple guards most jealously: the ability to design and ship beautiful, tightly integrated consumer hardware. Hire Ive's design shop, recruit hundreds of Apple's chip and on-device-AI engineers, and you are no longer a software partner. You are building a rival iPhone team from Apple's own bench. The lawsuit is Apple's attempt to raise the cost of that project — legally, reputationally, and operationally — before it produces a device.

The talent war's new phase

Read against the broader moment, the timing is telling. This is the same stretch in which top researchers have been changing labs like free agents — Noam Shazeer to OpenAI, John Jumper to Anthropic, a steady drain of senior names moving between the frontier houses for compensation packages that read like sports contracts. The industry has treated this churn as a cost of doing business, the natural friction of a field where a handful of people can move a roadmap by quarters.

Apple's suit is a bet that there's a line inside all that motion, and that OpenAI crossed it. Poaching an engineer is legal and, in this market, routine. Poaching an engineer and the confidential specifications in their head and on their laptop is not. The complaint is an attempt to draw that boundary in a venue that can enforce it — to convert a competitive grievance into a legal liability, and to put every AI lab currently hiring aggressively from its rivals on notice that the next aggressive hire might come with a subpoena attached.

What actually turns on this

The near-term legal outcome is almost beside the point. Trade-secret cases are slow, evidence-heavy, and frequently settle before a verdict. What matters more is the signal. Apple, a company famously reluctant to air its internal security in public, decided the threat was worth a lawsuit that will force it to describe exactly what OpenAI allegedly took — and therefore what Apple was building. That is a company that has concluded the normal tools of competition are no longer enough.

For OpenAI, the suit is a tax on its hardware ambition regardless of how it's resolved. Every future Apple hire now carries litigation risk. Every internal document touching io Products becomes discoverable. And the narrative — that OpenAI is building its consumer device on the back of Apple's stolen work — is now in the public record whether or not it's ever proven. The AI talent war spent two years as a story about who could pay the most for the best people. Apple just reframed it as a question of what those people were allowed to bring with them, and made it a matter for a judge.