Is OpenAI getting this wrong?

TL;DR: OpenAI fired back at Apple’s “you stole our hardware trade secrets” lawsuit accusations yesterday—by publishing private emails and messages. These don’t speak to the trade secrets issue, but instead paint Apple as shifting the blame for its supposedly sloppy lawyering and lax employee offboarding. But the jury’s (literally) still out on whether such sniping will help sway public opinion in OpenAI’s favor.

What happened: Yesterday, Apple asked the court for a preliminary injunction barring OpenAI from using or disclosing the company’s “proprietary and trade secret information.” OpenAI then clapped back by releasing some receipts.

In a blog post titled “Apple is getting this wrong,” OpenAI published emails from Apple’s outside legal counsel and iMessages from Chang Liu—a former Apple engineer and current OpenAI employee named in the suit Apple filed on July 10. What these messages seem to show is OpenAI’s version of events—that Apple’s lawsuit is “careless, aggressive and oddly personal” and that the AI lab does not have “nor want, any of their trade secrets.”

What’s in the emails: They reiterate what was made public weeks ago: that a lawyer representing Apple emailed the wrong person at OpenAI—then thanked him for a call the two never had. The point OpenAI seems to be making: Apple says it got no response from OpenAI about its trade secret concerns—but what actually happened was this email mix-up and then no follow-up from Apple to discuss the specific allegations in the lawsuit.

The iMessages: In its suit, Apple accuses Liu of “surreptitiously” downloading dozens of confidential files after leaving the company. These iMessages, though, show Apple employees asking Liu how to access files, including those from his old work iCloud. The point OpenAI seems to be making: Liu’s conduct wasn’t exactly surreptitious. Apple employees asked him for help, not necessarily the other way around. (Read the full iMessages, redactions and all, here.)

The he-said-she-said wars: These published messages don’t directly rebut the core allegation that OpenAI coordinated to steal Apple’s trade secrets. Instead, the AI lab seems to be making its opening arguments in the court of public opinion (which admittedly could have weight, especially if Apple is granted the jury trial it asked for). OpenAI has also aired dirty corporate laundry in a similar way before. In 2024, it released Elon Musk’s emails to fight a lawsuit he filed accusing the company of betraying its nonprofit founding mission. (Musk withdrew the suit, then filed a similar one a few months later.)

Bottom line: Selectively releasing receipts may not work the way OpenAI hopes. And if the preliminary injunction is granted, Apple could get access to OpenAI’s “devices, storage drives, and accounts”—potentially leading to more dirty laundry coming to light. —WK

Also at OpenAI…

• Add this to the pile of potential legal fights OpenAI could one day be embroiled in: being responsible for its AI agents’ autonomous hacks.