Apple Says Ex-Engineer Stole Trade Secrets, Then Helped OpenAI Destroy Evidence

Apple has accused a former engineer of stealing confidential trade secrets and then working with a colleague at OpenAI to destroy the evidence, according to new court filings in an ongoing lawsuit.

The dispute centers on Chang Liu, who left Apple to join OpenAI's hardware efforts. Liu's legal team recently handed over his old company-issued laptop for forensic investigation, and Apple says the analysis turned up evidence that Liu used - and later tried to hide - confidential company information.

Specifically, Apple alleges Liu downloaded a "highly confidential" Apple circuit schematic using his former work laptop, then uploaded it to LTspice, a widely used tool for simulating circuit performance before physical prototypes are built. Apple also claims a tool Liu used at OpenAI shares its name with an internal Apple engineering application, suggesting the overlap goes beyond a single document.

The more serious allegation, though, is what happened after Apple started investigating. According to the filing, once Liu learned in June that Apple was looking into his laptop use, he enlisted an OpenAI colleague, Yu-Ting Peng, to help destroy evidence - and Apple argues OpenAI itself was "well aware" Liu had access to its data.

OpenAI has pushed back, framing the dispute as a symptom of Apple's own offboarding failures rather than deliberate theft, and arguing that whatever files Liu accessed weren't actually relevant to the hardware work he's doing now.

The case is scheduled for arguments before a federal judge on October 1, and it lands at a sensitive moment for both companies. Apple has been racing to catch up on generative AI features, while OpenAI has been expanding aggressively into hardware - reportedly working on consumer devices with former Apple design chief Jony Ive. Trade secret disputes involving departing engineers are common in Silicon Valley, but the evidence-destruction allegation, if it holds up, could turn what might have been a routine dispute into something with real legal consequences for both the individuals involved and, potentially, OpenAI's own hardware ambitions.