OpenAI Hiring Records: Apple Trade Secrets Case 🚀 (2026)
Discover the OpenAI hiring records order in the Apple trade secrets case—what’s included, excluded, and why AI hardware device hiring interviews matter.
Apple's Discovery Win Puts OpenAI Hiring Under the Microscope
Apple has won expedited access to a specific set of OpenAI hiring materials in its trade secrets lawsuit, but the ruling is narrower than a broad request across OpenAI's hardware business. The court ordered OpenAI to turn over documents from interviews with former Apple employees who applied for roles in consumer hardware and robotics, and it approved access to private messages and depositions involving three people who previously worked at Apple.
That distinction matters. Apple says former employees took or shared confidential information as they moved to OpenAI. The judge's decision lets Apple examine how those candidates were interviewed and what was discussed. But it doesn't find that trade secrets were stolen, and it doesn't grant Apple unrestricted access to OpenAI's hardware research and development.
The timing also raises the stakes. OpenAI is building consumer hardware, so recruiting and information handling may be relevant to future products. Still, it's worth separating what a discovery order covers from what it can't prove.
What the judge ordered OpenAI to produce
U.S. District Judge Edward Davila granted Apple expedited access to materials tied to OpenAI interviews with former Apple employees who applied for roles in OpenAI's consumer hardware and robotics groups. Apple can seek documents such as interview checklists, guides, instructions, manuals, scripts, technical presentations, and records that show whether Apple technology or components came up during interviews.
The schedule is tight. OpenAI has five days to produce materials it has already collected and 14 days to produce other materials covered by the request. The court's deadlines reflect the decision that this particular discovery should move quickly - but they don't mean every document Apple requested is automatically due.
The order covers three kinds of evidence
The ruling isn't limited to interview documents, but it stays tied to specific people and topics within the case. Davila also approved Apple's access to private messages on LINE, a messaging app, between individuals named in the lawsuit, including OpenAI employees Chang Liu and Alyssa Peng.
The judge further approved depositions - sworn testimony - of Liu, Peng, and OpenAI chief hardware officer Tang Yew Tan. All three are former Apple employees. Depositions can help each side pin down what a witness remembers and what records support - or contradict - those memories. They're part of building the record, not a decision about whether anyone acted improperly.
What happens next depends on what the materials actually contain. A hiring guide could show the questions OpenAI used consistently. Interview notes could show whether candidates volunteered details, were asked about specific Apple products, or discussed confidential work. The messages could provide context about recruiting or information handling. But none of those documents, standing alone, automatically proves confidential material was misused.
Why the expedited timetable matters
Discovery can drag when the parties dispute what's relevant and what's too broad. Here, the expedited order changes the pace for the specific materials included in the request: the parties have less time to collect, review, and produce them.
That's useful when information is central to a quickly developing dispute. It also puts pressure on how the records are handled. Relevant material could be spread across email, interview systems, messaging apps, shared drives, and personal devices. Collecting it fast, while keeping context and protecting confidentiality, takes coordination.
Those five- and 14-day deadlines are for this order's scope, not a general timeline for the entire lawsuit. Other parts of the case may proceed under different schedules.
What Apple alleges - and what remains unproven
Apple filed its lawsuit in July and accused OpenAI of a pattern of trade secret theft. In the allegations described in the case, Apple claims former Apple employees used Apple laptops to download confidential files. Apple says some of those files involved unannounced technologies, features, and products, including technical specifications, engineering presentations, and proprietary project data.
Apple also alleges that former Apple employees at OpenAI coached people considering a move from Apple on what information to bring to OpenAI interviews. Interview materials matter to that claim because they can show what interviewers asked, what candidates presented, and whether discussions referenced Apple technology or components.
These are allegations from Apple, not findings created by the discovery order. When a judge allows a party access to evidence, it doesn't mean the evidence proves the accusation. It means the court has permitted a defined route for investigating the dispute.
The difference between discovery and a decision on the merits
In a lawsuit, discovery is the process for exchanging relevant evidence before trial or another resolution. Courts still have to decide whether a request is relevant, proportionate, and specific enough. Those rulings affect what can be collected, but they don't answer who's right on the underlying claims.
That difference can get lost in a high-profile technology case. "The court ordered documents produced" can sound like confirmation that the documents show wrongdoing. This order doesn't do that. It authorizes access to certain records and testimony while leaving key factual questions unresolved.
Those questions include whether confidential information was actually taken, whether it qualifies as a trade secret, whether it was used or disclosed, and who - if anyone - was responsible. The parties may also need to establish what employees knew, what they were allowed to access, and what happened to information after they left Apple.
The evidence has to be evaluated in context. A candidate describing experience working at a previous employer isn't automatically disclosing a trade secret. And a detailed technical presentation about unannounced work could raise questions if the material was confidential and obtained improperly. In the end, the underlying facts matter more than a headline or a single document seen in isolation.
Why hiring records matter in an AI hardware dispute
On the surface, interview checklists and recruiting records can seem like side material in a dispute about technology. But they can show how information moved between companies. This discovery ties together three steps that are often treated separately: access to confidential material at a former employer, the interview process at a new employer, and subsequent work on products or research.
For Apple, the interview materials could help test whether candidates were prompted to discuss Apple's technology or whether they brought those details independently. For OpenAI, the same records could shed light on what its interview process asked candidates to talk about - and whether it was set up to avoid confidential information. Either way, the records may support or undermine the competing accounts, depending on their contents and the other evidence in the case.
The hardware context raises the stakes, but it doesn't change the legal standard
OpenAI is developing a consumer hardware device. If it becomes a widely used way for people to access software and services, it could place OpenAI in more direct competition with established device makers, including Apple. That's one reason hardware expertise and recruiting show up in the public discussion around the lawsuit.
But commercial rivalry doesn't prove that a trade secret was taken. Hiring experienced people from a competitor doesn't, by itself, prove wrongdoing either. The legal and practical issues come down to specific information: whether it was protected, how it was obtained, whether it was shared, and whether it was used in a way the law prohibits.
This is also why the case has broader implications. Tech companies hire experienced people all the time, and they need clear boundaries between general expertise and confidential information tied to a former employer. Interview processes are one place those boundaries can either be documented clearly or become hard to reconstruct later.
A careful process doesn't require interviewers to avoid all discussion of a candidate's prior work. It does require attention to confidential projects, unpublished specifications, source material, customer data, and anything a candidate isn't authorized to share. Clear instructions and consistent record-keeping can help both the hiring company and the candidate understand where the line is.
Why process records can turn into evidence
Recruiting records are created for practical reasons: to standardize interviews, compare candidates, and document hiring decisions. When a dispute arises, those same records can show what the company's process looked like at the time.
For instance, a consistent interview guide can support the idea that candidates were evaluated against a common skills list rather than being asked to disclose a former employer's confidential plans. An interview note or presentation that references a particular product may prompt questions about where the information came from and how detailed it was. The point isn't that any single type of record automatically favors one side. It's that contemporaneous documentation is often more reliable than memories formed after a lawsuit begins.
The broader lesson is pretty direct: hiring teams working in sensitive technical areas need to know what interviewers can ask, what candidates shouldn't bring, and how to handle documents or presentations that appear to include confidential material. Written guidance matters, but it only helps if interviewers follow it and the company can show what happened in a specific interview.
What the ruling does not give Apple
Davila did not approve all of Apple's requests for expedited access. He denied Apple's request to obtain OpenAI research and development materials in specific areas, including metal finishes and power management.
The judge also denied several broader discovery requests, including requests for access to OpenAI's hardware research and for information about what former Apple employees created while working at OpenAI. Those denials narrow what Apple can inspect right away. They also act as a counterweight to the idea - often repeated in headlines - that OpenAI must turn over a wide range of internal technical work.
A denial is not necessarily the final word
The judge left open the possibility that Apple could narrow its requests and seek access again. That means the court hasn't ruled that the broader topics are automatically irrelevant. Instead, the specific requests submitted for expedited treatment weren't granted as written, and a more focused request could be considered later.
Discovery disputes often work this way: a broad request can sweep in large amounts of sensitive material, including content that doesn't connect closely to the claims. A narrower request can target specific projects, people, time periods, or categories of documents and make the connection to the dispute clearer.
For OpenAI, a narrower scope still means dealing with additional collection and review if Apple successfully renews its request. For Apple, refining the requests may require explaining more precisely how particular research or work product relates to the allegations. In practice, that's not the end of the fight - it's a continuation over relevance and scope rather than an open-ended right to inspect a competitor's technical work.
What to watch next
The immediate next step is OpenAI's production of the materials covered by the deadlines. After that, both sides can use documents and depositions to develop their accounts, challenge interpretations, and potentially seek further requests. Apple may also revise some of the requests the judge declined to expedite.
As readers track the case, it helps to separate three types of developments:
- Evidence production: documents or messages are collected and shared under the court's order.
- New allegations or responses: a party argues what the evidence means.
- A judicial finding: the court decides a legal or factual issue.
This ruling falls under the first category. It authorizes access to specific evidence. It doesn't settle the lawsuit, prove that Apple's allegations are correct, or determine whether OpenAI's hardware work used Apple trade secrets.
Conclusion: A targeted discovery win, not a finding of wrongdoing
Apple's latest court win gives it expedited access to interview records, certain LINE messages, and testimony from three former Apple employees now at OpenAI. The evidence could help clarify what was discussed during hiring and how information moved between the companies. At the same time, the judge rejected expedited access to broader categories of OpenAI hardware research and left Apple to narrow some requests if it wants to try again.
The most useful way to read the decision is as a ruling about evidence and scope, not as a verdict on the trade secrets allegations. What matters next is what the records show, how the parties interpret them, and whether the court approves any narrower follow-up requests.
For the original report and further case details, read TechCrunch's coverage of the ruling.