Apple Claims Former Employee Trained OpenAI Bot on Apple Secrets
In a fresh court filing submitted to a California federal court, Apple has brought forward what it calls “shocking evidence” showing that former employees stole trade secrets and actively worked to cover up their tracks while building hardware for OpenAI (via MacRumors).
The drama centers on Chang Liu, a former senior electrical engineer who spent eight years at Apple before departing for OpenAI in January. According to Apple’s latest court documents, forensic teams examined an Apple-issued laptop that Liu retained after leaving the company.
Apple claims that Liu did not just take confidential hardware details when he walked out the door. He actually put them to work inside OpenAI’s offices. The legal documents reveal that Liu downloaded a confidential Apple circuit schematic two months after leaving the iPhone maker. He then loaded that secret diagram into LTspice, a specialized engineering simulation program, to perform power conversion development work for OpenAI.
Even more concerning for Apple is how Liu automated the process. In internal messages uncovered by investigators, Liu bragged to colleagues about training an artificial intelligence agent to read the stolen schematics.
“In the past hour, my agent learned how to run LTspice, look at result, tune compensation parameter,” Liu wrote in one message cited in the filing. He added that he was “feeling AI all day long” while setting up the automated system.
Apple attorneys warned the judge that training AI models on stolen proprietary data creates an “irreversible and continually propagating” security nightmare, as the underlying neural networks absorb Apple’s trade secrets into their long-term training memory.
The new evidence goes far beyond simple data downloading. Apple alleges that Liu and other OpenAI team members were fully aware that he still had back-door access to Apple’s cloud storage system weeks after his resignation. When Liu realized Apple’s corporate security team had opened an internal investigation into his activities, he did not turn himself in. Instead, he fired off text messages to a colleague at OpenAI, giving her explicit instructions on how to destroy digital evidence.
A formal hearing on Apple’s request for fast-tracked discovery is set for October 1.
Want to see more of our stories on Google?
P.S. Want to keep this site truly independent? Support us by buying us a beer, treating us to a coffee, or shopping through Amazon here. Links in this post are affiliate links, so we earn a tiny commission at no charge to you. Thanks for supporting independent Canadian media!
