In a 31 page filing, OpenAI argues Apple's complaint names broad 'product development categories' instead of specific secrets — a pleading standard gap that could end the case before any evidence is examined.
OpenAI's 31-page motion to dismiss Apple's trade-secrets lawsuit is the first real legal test of a fight that, until Thursday, had lived mostly in press releases and hiring announcements. Filed in U.S. District Court for the Northern District of California on August 6, the motion does not deny that two former Apple engineers joined OpenAI's hardware effort. It argues that Apple has not done the basic paperwork required to keep the case alive.
Under the federal pleading standard set by Bell Atlantic v. Twombly and Ashcroft v. Iqbal, a plaintiff must describe the alleged wrongdoing with enough detail that a defendant can actually defend itself. Vague labels like "confidential information" do not count. OpenAI's filing leans on that standard hard, calling Apple's "generalized product development categories" insufficient to survive a motion to dismiss.
"Apple's complaint fails to sufficiently describe the information for which it claims trade secret protection," OpenAI's lawyers wrote, per the filing on CourtListener (doc 59). The motion repeats that argument, by Axios's count via American Bazaar, nearly fifty times: a word similar to "fail" appearing on almost every other page.
A motion to dismiss is decided on the four corners of the complaint. OpenAI does not have to prove the secrets were not taken; Apple has to describe them well enough for OpenAI to defend. If the court agrees, the trade-secrets claims are gone before either side takes a deposition. If it disagrees, the case moves to discovery, where Apple's pending preliminary injunction request could force OpenAI to wall off any work that touches on what Apple claims it stole.
The filing also pushes back on the underlying theory of theft, arguing that OpenAI is "building something entirely new" and has "no use, need or desire" for Apple's secrets. And it raises a practical defense: the two former engineers, OpenAI says, used personal iCloud accounts for some Apple work, so the personal-and-work mix makes it hard for Apple to show they knowingly took confidential material (AppleInsider, MacRumors).
The motion's most quotable line is a counter-charge: that Apple is suing because it cannot compete on AI integration and cannot keep the engineers it wants to keep. "Apple cannot transform the foreseeable consequences of its own systems and policies into evidence of theft by former employees," the filing argues, per the CourtListener docket.
The filing also surfaces a public-record gaffe that Apple's outside counsel would probably rather not have in the docket. According to OpenAI's account, when Apple first tried to contact OpenAI about the alleged theft in February, its outside lawyers emailed the wrong person, mixing up two Asian last names, and only later admitted the error. PYMNTS and The Next Web both report the mix-up as a factual exchange, not a punchline. The detail matters because it undercuts Apple's narrative that OpenAI was ducking a serious outreach.
What happens next is a procedural fork. Apple's preliminary injunction request, filed earlier this week, asks the court to bar OpenAI from using any of the alleged secrets during the case. The motion to dismiss, if granted, makes that hearing unnecessary. If denied, the injunction fight becomes the next public stage, and OpenAI's hardware work becomes the subject of an evidentiary fight over what its engineers knew and when.
For now, the case is being fought in court and in the press at the same time. OpenAI's filing does both jobs in one document: it tests the legal floor of Apple's complaint, and it tells a story about why Apple filed the suit in the first place. The judge gets the first word on the law. The market will be reading the rest.