Apple says a forensic laptop exam caught a former engineer destroying evidence in its OpenAI trade-secrets suit

Apple told a federal court on August 31 that forensic analysis of a former engineer's company-issued laptop turned up what it called "shocking evidence": that the engineer, Chang Liu, downloaded a confidential Apple circuit schematic two months after leaving for OpenAI, then asked an OpenAI colleague to help destroy it once he learned Apple was investigating him. The filing is a new evidentiary chapter in a trade-secrets lawsuit Apple brought on July 10, and it is the most specific, most technical claim Apple has put in front of a judge so far.

The July complaint names four defendants: Liu, an eight-year Apple system electrical engineer who left for OpenAI in January 2026; Tang Tan, a 24-year Apple veteran and former VP of iPhone and Apple Watch product design who departed in February 2024; OpenAI itself; and io Products, the hardware studio OpenAI absorbed in a $6.5 billion deal for Jony Ive's design team. Apple's own framing of the case, quoted directly from its complaint: "This case is about Apple's former employees stealing Apple's trade secrets for the benefit of OpenAI." The original filing also alleged Tan used confidential Apple project codenames in OpenAI job interviews, directed Apple staff to bring real hardware components to OpenAI "show and tell" sessions, and that Liu separately exploited a security gap to download a compilation of manufacturing files running over a thousand pages.

The new filing's central claim is narrower and more concrete than the July complaint's broader culture-poaching allegations. Apple says Liu used the schematic to run simulations in LTspice, an electrical-engineering circuit-simulation tool, and that he trained an AI coding agent to operate that software for him -- telling a colleague, according to Apple's filing, that "in the past hour, my AI agent learned how to run LTspice, look at result, tune compensation parameter," cutting a task he said used to take a full day down to two hours. Apple also says Liu and others used an OpenAI internal tool that shares its name with an Apple engineering application, and that Liu accessed Apple's third-party cloud storage without authorization after he left.

"The MacBook represents the very limited information Defendants provided so far (and only after weeks of delay), and shows Apple is not conducting 'fishing expeditions' but that its trade secrets are being used and evidence is being destroyed."

That quote, from Apple's own filing, points at the evidence-destruction allegation specifically: that once Liu learned of Apple's investigation, he instructed OpenAI colleague Yu-Ting Peng to "restore" his Apple-issued devices -- Apple's word for what it characterizes as an attempt to erase forensic evidence -- and that Peng confirmed she would do it. OpenAI has not, in anything reviewed for this piece, directly addressed that specific claim. Its response instead reframes the dispute at a system level.

OpenAI's defense is, in effect, an argument about Apple's own IT hygiene: that Apple lets employees sync work files through personal iCloud accounts, and that its practice of immediately escorting departing staff off the premises leaves no clean process for separating a person's own devices and data from the company's. Neither point, if true, answers the more specific allegation -- that Liu kept using a specific confidential schematic for OpenAI's benefit, or that he asked someone to help erase it once Apple came looking. Those are the two claims the October hearing will actually turn on.

The 400-plus figure is doing real work in Apple's filing, even though it names no specific wrongdoing by most of those people: it's the backdrop Apple wants the judge to see this case against, not two engineers but a large and ongoing flow of hardware talent from Apple to a company that, as of this year, builds physical devices of its own. That framing is also why the suit reads differently from most AI-industry trade-secret disputes, which are usually about model weights or training data. This one is about circuit schematics, manufacturing files and the kind of hands-on hardware-engineering know-how that took Apple two decades to build inside the iPhone supply chain -- and it lands at the same moment OpenAI, through io, is trying to ship its own first consumer hardware product.

There's also a smaller, almost self-referential detail buried in Apple's own filing: part of what it says Liu did with the schematic was hand it to an AI coding agent. According to Apple's account, Liu told a colleague his agent "learned how to run LTspice, look at result, tune compensation parameter" -- cutting a task from a full day to two hours. If Apple's version holds up, the case isn't just about a document leaving the building; it's an early example of agentic AI tooling compressing the time it takes to turn a stolen schematic into working engineering output, which is precisely the kind of leverage that makes a 20-year-old chip-design headstart harder to protect than it used to be.

What Apple is asking the court to do also matters beyond this one case. The preliminary injunction it wants would block OpenAI hardware work that relies on Apple's technology -- a request aimed squarely at io, the design studio built around Jony Ive that OpenAI is counting on for its first consumer device. A judge granting that motion wouldn't just resolve a dispute between two engineers and their former employer; it would put a court-ordered speed bump in front of OpenAI's hardware timeline while the underlying trade-secret claims are still being litigated. That's the real reason this case is worth tracking past the headline evidence-destruction claim: the outcome touches a product OpenAI hasn't shipped yet.