Apple filed suit against OpenAI on July 10, claiming engineers who left the company stole trade secrets to advance the AI startup’s hardware plans. The lawsuit names OpenAI, IO Products (Jony Ive’s hardware company acquired by OpenAI in 2025), and two former Apple employees: Tang Tan (24 years at Apple, now chief hardware officer at OpenAI) and Chang Liu (eight years on iPhone, joined OpenAI in January 2026).

Apple contends it discovered «a pattern of theft» that includes access to company systems weeks after Liu’s departure, downloads of dozens of confidential files, and instructions to colleagues on how to copy documents and «avoid trouble» with security. During interviews for OpenAI, according to the complaint, Apple candidates were asked to bring «components they were working on» and samples of unreleased products.

Why this matters to you. This case illustrates two things: the talent war between big tech and AI labs has escalated to lawsuits that question what an employee can take when switching companies, and Apple is building a documented case around specific access logs and files, not just generic confidentiality breaches. If you work in a field where engineers move between competing companies, the precedent from this lawsuit will matter: what can be discussed at an interview, what stays bound even after you leave, how strict post-employment controls become.

OpenAI responded that it has «no interest in others’ trade secrets,» but the trial will focus on the facts, not the tone.

In detail

The context missing from announcements

The lawsuit arrives at a precise moment: OpenAI announced for 2027 its first in-house AI hardware device, following the 2025 acquisition of IO Products. Tang Tan, one of the three defendants, leads that project. Before leaving Apple in 2024, Tan was vice president for Apple Watch: 24 years of career, access to confidential projects, direct knowledge of manufacturing partners.

The second defendant, Chang Liu, spent eight years on iPhone as a systems electrical engineer. Apple accuses him of retaining «at least one company computer» after departure, of not signing the confidentiality reminder or scheduling an exit interview. Weeks after his last day, Liu allegedly discovered an authentication vulnerability that gave him continued access to Apple’s network storage. In a chat with Yu-Ting Peng (the third named person), Liu writes: «LOL, I found I can access the [network storage], so funny». Peng responds: «I’m ready».

Apple contends that Liu downloaded «dozens of confidential files,» including technical specifications, details of unreleased products, engineering presentations, and proprietary project data. The complaint also cites Liu telling Peng, before she joined OpenAI, to use Line Messenger to communicate and avoid detection.

The most concrete accusations

The complaint contains six points Apple considers its «strongest claims,» all backed by message excerpts, dates, and file names. First: Apple candidates at OpenAI interviews invited to bring «CAD/design artifacts» and physical prototypes. Second: Tang Tan emailing himself information about Apple suppliers before departure. Third: Liu instructing a colleague how to copy confidential files before she follows him.

Apple also accuses OpenAI of directly contacting an industrial partner that works with Apple on design and metal fabrication, asking it to apply proprietary techniques developed with Apple to OpenAI projects. This part of the suit matters because it expands the accusation beyond individual employees: if confirmed, it would show a systematic effort to replicate not just internal knowledge, but the supply chain itself.

Where sources align and where they diverge

The Verge and TechCrunch agree on the main facts: documented post-employment access, downloaded files, components brought to interviews. TechCrunch emphasizes that some messages between Liu and Peng joke about unauthorized access («LOL»), a detail that in court will weigh on awareness. The Verge focuses more on OpenAI’s strategy of building hardware by replicating «the secret technologies, business processes, and supply chain innovations that took Apple decades to develop».

What no source has yet verified: whether OpenAI knew of Liu’s unauthorized access before the lawsuit, and whether the downloaded files were actually used in developing the announced hardware device. OpenAI’s response is generic and doesn’t address the specific evidence.

The practical implications

If Apple wins, the precedent will make it costlier for AI labs to hire from incumbents with large IP portfolios. If it loses, or if the case settles under confidentiality terms, the signal goes the other way: aggressive talent poaching works, and post-employment controls are hard to enforce.

For those building with AI, this story isn’t just courtroom gossip: it’s a signal about how tight controls will become on using knowledge from previous jobs when projects involve models competing with major platforms. And about how to document where your knowledge came from, if you ever have to defend yourself against a similar accusation.