Judge orders OpenAI to turn over hiring records in Apple trade secrets case

At the heart of this lawsuit is Apple's allegation that OpenAI engaged in a pattern of trade secret theft. In July 2026, Apple formally sued OpenAI. Three months later, in October, U.S. District Judge Edward Davila ruled on Apple's motion for expedited discovery: partially granted, partially denied.

The granted portion centers on the recruiting process. The court allowed Apple to view documents from OpenAI interviews of former Apple employees who applied for positions on its consumer hardware and robotics teams. The materials may include interview checklists, guides, instructions, manuals, scripts, as well as technical demonstrations and records showing that interviews discussed Apple technology or components.

The denied portion points at R&D itself. The judge denied Apple's request to expedite access to OpenAI's R&D information in specific areas such as metal surface treatment and power management, and also denied several broader discovery requests, such as obtaining information about OpenAI's hardware research or content created by its former employees during their time at OpenAI.

This division itself sends a signal: the court is willing to let Apple investigate whether improper information flows occurred in the recruiting process, but is unwilling to let Apple cut directly into OpenAI's R&D core at this early stage of the case.

The court set clear deadlines while granting expedited discovery. OpenAI must hand over already-collected materials within 5 days and other materials related to this request within 14 days. The judge agreed with Apple's position that these documents should be handled on an expedited basis.

The distinction between the two deadlines is worth noting: the 5 days applies to materials OpenAI has already collected, while the 14 days is left for the portion not yet assembled. This means the court struck a compromise between efficiency and feasibility, rather than imposing a one-size-fits-all requirement that all materials be produced within an extremely short time.

But the deadlines themselves do not guarantee compliance. Whether OpenAI will fully produce the materials within the deadlines is not confirmed anywhere in the source and remains unknown.

In addition to recruiting documents, the judge also ruled that Apple may obtain private LINE messages between named individuals in the case, including OpenAI employees Chang Liu and Alyssa Peng. At the same time, the judge approved depositions of Liu, Peng and OpenAI Chief Hardware Officer Tang Yew Tan. All three are former Apple employees.

The simultaneous approval of private messages and testimony means the court is allowing Apple to cross-verify from two directions: first, what information was actually conveyed during the recruiting process, and second, how the relevant individuals describe that information under oath.

However, the specific content of the named LINE messages and their relevance to the case are not disclosed in the source. Whether these messages can support Apple's allegations cannot currently be determined.

In the lawsuit, Apple alleges that former employees used Apple laptops to download confidential materials, as well as files containing unpublished technology, feature and product information, including technical specifications, engineering demonstrations and proprietary project data. Apple also claims that OpenAI's former Apple employees coached others considering leaving Apple to join OpenAI on what information to bring to interviews.

These are allegations made unilaterally by Apple and have not been confirmed by a court ruling. It is necessary to distinguish: the court granting discovery does not equal the court endorsing the allegations. The discovery order only shows that the court believes Apple is entitled to obtain these materials to advance the litigation, not that it has made a judgment on the substantive dispute.

From the content of the allegations, Apple is trying to construct a chain from "downloading confidential material" to "bringing it to interviews" to "using it after joining." But whether this chain holds depends on subsequent evidence and cannot currently be predicted.

This case arises as OpenAI is developing a consumer hardware device. The source says that if this AI device becomes a core part of consumers' lives, it could in the future compete with the iPhone or smartphones. The relevant interview materials concern recruiting activity before the launch of OpenAI's rumored first hardware device.

It must be made clear that the launch timing and form of OpenAI's first hardware device are only rumored in the source and have not been confirmed. Directly linking the lawsuit to hardware competition is an inference based on overlapping timing, not an established fact.

But this background explains why Apple is so sensitive about the recruiting process: if OpenAI's hardware team is recruiting heavily from Apple, then whether the interview process involved Apple technical information becomes a question Apple must investigate.

The judge's ruling denying Apple's broader discovery requests is not a final ruling, and the judge preserved room for Apple to try again after narrowing the scope of its requests. This means the R&D information currently shut out could still enter the discovery scope in the future through more precise requests.

For Apple, the strategic choice becomes clear: rather than insisting on broad R&D discovery requests that are repeatedly denied, it would be better to focus on the already-approved recruiting materials and personnel testimony, and look for specific leads sufficient to support the next round of requests.

For OpenAI, the 5-day and 14-day deadlines are the immediate compliance pressure, but the longer-term variable is this: if Apple finds evidence in the recruiting materials pointing to the flow of R&D information, the currently denied requests may be reintroduced in a new form.