iCloud habits become OpenAI’s defense in Apple suit

OpenAI now argues that Apple never guarded its hardware secrets well enough for them to count as trade secrets. Its rebuttal filing blames Apple’s own offboarding: Apple encourages staff to use personal iCloud accounts and gives departing employees no time to return company devices, transfer internal files back or hand over their work, AppleInsider reported on September 1.
Per Reuters, the filing again accuses Apple of trying to stop people from leaving, and this time also takes a swipe at Apple’s AI record. ”[Employees] can leave a company like Apple that has struggled to adopt AI and move to an exciting startup that builds innovative products,” OpenAI wrote. “Apple may not like those choices. But it cannot claim those choices are unlawful, and it cannot use its own sloppy procedures to blame others for its own mess.” It also says Apple “is trying to blame everyone else” for its own failings.
From those procedures OpenAI draws two consequences: employees find it hard to separate personal iCloud data from company data, and ex-employees end up keeping company devices. Apple sued on July 10 in federal court in Northern California, alleging, per CNBC, that “at every level, from members of its Technical Staff to its Chief Hardware Officer, and in coordination with business partners, OpenAI has been stealing Apple’s trade secrets and confidential information.” Apple says former engineer Chang Liu never returned his Apple-issued laptop, skipped his exit interview, did not sign a confidentiality reminder and downloaded dozens of confidential hardware files. Apple has since filed an initial forensic analysis of Liu’s MacBook, which it says found an Apple circuit schematic used in his work at OpenAI.
Apple also alleges that Tang Tan, OpenAI’s hardware chief and a former Apple vice president, circulated an internal Apple document marked “Need to Know” that described Apple’s exit security procedures, so new hires could avoid scrutiny when they left, and advised recruits not to tell Apple they were leaving for OpenAI. AppleInsider’s own reading is blunt: what has been reported of the new filing so far, it wrote, reads as “tantamount to admitting guilt.”
Information only qualifies as a trade secret if its owner took reasonable steps to keep it secret, so showing Apple was careless could knock out the claim rather than merely shift blame. A defense attorney who handles trade-secret cases told AppleInsider in a brief call on Tuesday that the bar for proving the owner failed to protect its data is very high, and that on the public record Apple should have no trouble clearing it.
On OpenAI’s specific points, the attorney said precedent treats a very short window for returning devices as acceptable, sometimes only hours, and Apple seems to meet that, especially given its allegation that the departing engineers deliberately skipped check-out interviews. If an employee is proven to have dodged check-out and turn-in, the company is generally given the benefit of the doubt, and the blame rests with the employee who allegedly misused the data. The cloud argument has historically been hard to make too: precedent generally puts the responsibility for handling data ethically on the account holder, here the ex-employee, particularly if exit interviews were avoided.
Then there is the change of position. When the suit was filed, an OpenAI representative said, per CNBC: “We have no interest in other companies’ trade secrets. We remain focused on building innovative technology that empowers people everywhere.” The attorney said that moving from denial to a tacit admission only after Apple’s initial forensic filings will be damning. The next hearing is set for early October.
“Judges have long memories,” the attorney said. “When Apple filed, OpenAI already knew if they did or didn’t do it, and should have gone this route from the beginning.” The shift, in their words, “looks like Mom busted the kid on the ladder, with his hand in the cookie jar, and crumbs everywhere.” And: “Apple’s consistency in accusation will be clear, and OpenAI trying anything that will stick will be messy.”
Up next
3.1% word error rate leads the streaming speech chart, Video
$0.18 an hour buys Meta’s new streaming transcription model, Muse Voice Transcribe, which logs a 3.1% word error rate ahead of GPT Live Transcribe and Gemini 3.5 Transcribe Live. The weights stay closed.
Related stories
- Judge keeps Apple’s X Corp settlement away from OpenAI
- MacBook analysis adds evidence to Apple’s OpenAI suit
- Texas sues OpenAI and Sam Altman over ChatGPT safety
- Eight states ask courts to pause OpenAI’s self-improving AI
- Florida wants outside oversight before OpenAI’s next model
- OpenAI’s 2019 note: “We trained GPT-3 on pirated stuff”
Comments
No comments yet. Be the first.
Join the conversation
Sign in to leave a comment. It appears after moderation.

