A trade-secret fight between Apple and OpenAI over purloined chip designs now hinges on a forensic examination of an engineer’s old work laptop.
Fresh filings accuse OpenAI of two things: letting a former engineer keep using confidential circuit material after defecting, and failing to stop an attempt to bury the evidence once he knew he was caught.
The engineer is Chang Liu, who left Apple for OpenAI in January. His former employer sued on July 10, and his lawyers surrendered his Apple-issued MacBook only on August 21.
Examination dated the download of a power-converter circuit design, along with simulation inputs, to March 7 from Apple’s cloud. By March 18 the files were feeding LTspice under Liu’s profile; related outputs later crossed his iCloud from a Mac Mini and reached the laptop in April.
Apple further points to messages in which Liu described an AI agent he had trained to operate LTspice, review outcomes and adjust compensation values, with jobs that once consumed a full day dropping to two hours. Its brief argues such learning spreads a secret endlessly, since models retain what they absorb.
Apple’s obstruction theory centers on June, when it says Liu, newly aware that investigators were closing in, talked with OpenAI colleague Yu-Ting Peng about erasing material Apple would need. The company adds that OpenAI knew he retained access through a hole in its authentication that had never surfaced.
OpenAI calls the case meritless, blaming Apple’s account administration for any lingering access. Apple’s wish list for the judge includes putting its designs off limits to OpenAI product work for the duration and compressing the evidence calendar, noting that hundreds of its alumni now sit at OpenAI and the full extent of the leak is still unknown.
