Brian Reed: Hey — quick question before we get into the actual legal filings. Have you ever seen a company publish a defendant's iMessages and emails the morning before their lawyers file the motion?
Eliza Ward: Not as a standard move, no.
Brian Reed: Because that's what OpenAI did yesterday — August 4, blog post called 'Apple is getting this wrong,' sharing private message exchanges. And then the 31-page motion to dismiss lands August 5.
Eliza Ward: Hold on — so the blog characterizes Apple's complaint as 'careless, aggressive, and oddly personal.' That language is doing something specific. They're not — I mean, that's not how you describe a complaint you think is going to get summarily dismissed.
Brian Reed: Unless you want to make sure the dismissal sticks in people's heads as a vindication, not just a procedural outcome.
Eliza Ward: Okay, but let me back up to the confirmed fact. Apple filed this lawsuit July 10 — U.S. District Court, Northern District of California — against OpenAI, Chang Liu, and Tang Tan. Trade secret misappropriation. That's the actual claim.
Brian Reed: And OpenAI's response, in the filing itself, is that they have 'no use, need or desire for Apple's trade secrets.' Which — hang on, is that a denial, or is that just saying Apple didn't specifically identify what was taken?
Eliza Ward: Those are different arguments, and the motion runs both of them. That's actually the thing I want to pull apart.
Brian Reed: Right — but the part that breaks the 'poached engineer' framing is the actual sequence. Tang Tan didn't leave Apple and walk into OpenAI. He left Apple, co-founded a hardware startup called io Products, and then *that company* merged with OpenAI in July 2025.
Eliza Ward: Which is — wait, that's a year before Apple even filed suit.
Brian Reed: Exactly a year before, yeah. So Apple's framing — 'systematic recruitment to steal secrets' — that framing has to survive the fact that OpenAI didn't recruit Tan. They acquired a company he built.
Eliza Ward: And now he's OpenAI's chief hardware officer. Twenty-four, twenty-five years at Apple overseeing iPhone and Apple Watch product design, then io Products, then this. That's the spine of Apple's whole hardware-espionage theory — not Chang Liu, who's named in the suit as a former senior systems engineer on iPhone but is, I think, secondary to the Tan narrative.
Brian Reed: The part I don't get — legally, does the acquisition route change anything? Like, if the secrets traveled through io Products and then into OpenAI via the merger, is that cleaner for OpenAI than a direct hire?
Eliza Ward: That's actually — okay, here's the plain version. Trade secret law is basically: imagine a master chef leaves a restaurant. Takes his knife skills, his palate, everything in his head — that's fine. But if he walked out with the written recipe for the house sauce, that's the violation. The law says you can take your expertise; you can't take the document. And every one of these cases is a fight over which one you actually walked out with.
Brian Reed: And Apple's argument is that Tan — and Liu — walked out with the recipe.
Eliza Ward: Right. And OpenAI's motion says Apple never actually specified *which* recipe. 'Product development designs' is — a judge will read that and ask, designs for what, exactly, and how do you know they ended up in ChatGPT hardware. Apple's allegation is that OpenAI wanted Apple's secrets specifically to build hardware for ChatGPT. That's a concrete claim. But specifying *what* crossed the line is what the motion to dismiss is hammering.
Brian Reed: So the distinction between 'startup acquisition' and 'direct poach' might not actually protect OpenAI — it just complicates how Apple has to prove the recipe moved.
Eliza Ward: Right — but complicating the acquisition route is actually the popular take I want to push back on. The one circulating right now is basically: four hundred former Apple employees at OpenAI, that's the conspiracy, case closed. And that number is doing so much work.
Brian Reed: Because four hundred sounds like a coordinated operation.
Eliza Ward: It does. But headcount is not a trade secret. Rule 12(b)(6) doesn't ask how many people left — it asks whether Apple specifically identified the confidential information those people allegedly took. 'Product development designs' for iPhone and Apple Watch is a category, not a specification.
Brian Reed: So the popular take gets this backwards — more people doesn't mean more proof.
Eliza Ward: That's — yeah, exactly that. Volume and circumstance versus legally specific misappropriation. Apple hasn't bridged that gap yet in the complaint.
Brian Reed: And then OpenAI piles on with the wrong-email story — outside counsel emails the wrong person, falsely claims a call with OpenAI's general counsel that apparently never happened. That's not about the substance, that's just — I mean, they're using that to build a 'careless' frame around the whole complaint.
Eliza Ward: Which is procedural ammunition. Separate from whether anything was actually taken.
Brian Reed: And here's the slippage that actually bothers me — OpenAI's filing also recharacterizes what the named employees were doing. Their version is those people were 'trying to help Apple, at Apple's request, to make sure their former colleagues' work could continue without them.' That's not 'we didn't take anything.' That's 'what you're calling theft we're calling continuity.' Those are completely different defenses running in the same document.
Eliza Ward: Which — wait, that gap between denial and pleading deficiency, that's actually what shapes the preliminary injunction fight. And that's the motion we haven't gotten to yet — the one that could stop OpenAI's hardware program before this case resolves at all.
Brian Reed: That gap is the whole ballgame right now. Because Apple didn't just file for dismissal — they filed a preliminary injunction. Which means there are two motions sitting in front of the same Northern District judge at the same time.
Eliza Ward: And the sequencing is everything. If the judge rules on OpenAI's motion to dismiss first and grants it — the injunction is moot. Case over, hardware program moves forward unconstrained.
Brian Reed: But if the injunction gets addressed first—
Eliza Ward: Then Apple can potentially freeze OpenAI's hardware development before the merits ever get litigated. That's not a years-away outcome. That could be — I mean, picture an engineer in OpenAI's hardware lab right now, working on a device spec that references display tolerances or sensor integration from, let's say, a wearable form factor. The injunction motion is the thing that could stop that work tomorrow. Not some trial in 2028.
Brian Reed: So the judge's sequencing decision — which motion to address first — that's actually more consequential than either motion's outcome on its own?
Eliza Ward: That's — yeah, I think that's the right frame. A granted injunction gives Apple enormous leverage before OpenAI even gets to challenge the substance. Tang Tan is OpenAI's chief hardware officer. If his work gets enjoined, that's not a legal inconvenience — that's the hardware program stalled at its leadership.
Brian Reed: And we don't know yet which motion the judge picks up first. That's genuinely unresolved.
Eliza Ward: Confirmed unknown, yeah. But — wait, there's a second layer. However this resolves, it's in the Northern District of California. This is going to be the case that gets cited every time an AI company hires from a competitor going forward. The trade secret specificity question, the acquisition-versus-poaching distinction, all of it.
Brian Reed: So the thing to actually watch isn't the blog post, isn't the 400-employee number — it's which motion the judge addresses first, and whether Apple can get specific enough on what crossed the line to survive past the pleading stage.
Eliza Ward: The blog post from August 4 — one day before the filing. And I genuinely don't know whether that's 'we're so confident in these iMessages that sunlight helps us,' or whether it's — actually, I'm not sure those two things are mutually exclusive.
Brian Reed: Let me see if I can make that concrete. If OpenAI's legal team thought the motion to dismiss was airtight on its own, you probably don't need the press. The specificity problem Apple has — 'product development designs' — that's a pleading deficiency a judge finds without public pressure. So why go loud the day before?
Eliza Ward: Unless the motion to dismiss succeeds and the preliminary injunction doesn't. Which is — I mean, those travel separately. You can win on 12(b)(6) grounds and still face an injunction argument. Maybe the blog post is for that fight, not the dismissal.
Brian Reed: I can't land on either side of that either. It's either confidence or anxiety and the same evidence fits both.
Eliza Ward: Yeah. I don't have a clean answer. That question's still open.
