Is the Apple-OpenAI Lawsuit Really About Trade Secrets, or About Who Controls What Comes After the iPhone?
A lawsuit dressed up as corporate espionage is really a fight over who owns the next twenty years of computing. Here's my read on it, plus the trade secret law behind the case, explained plainly.
TL;DR: Apple sued OpenAI, its hardware unit io Products, and two former Apple employees on July 10 for trade secret theft. The complaint reads like a thriller: a forgotten laptop, a “LOL” text, a Chief Hardware Officer accused of running the operation. The more interesting story is what happens eighteen months before a lawsuit like this gets filed, and why California’s unusual employment law makes trade secret claims the sharpest tool a company like Apple has left. Below: what the complaint actually alleges, a plain-English primer on how trade secret law works and why it exists, and my own opinion on who’s right (I’m Team Apple, with caveats). Plus what this signals for anyone whose business depends on a platform relationship that could sour.
Apple filed a 40-page lawsuit against OpenAI on July 10. Most of the coverage stayed on the parts that read like a thriller: a forgotten laptop, a text message that opens with “LOL,” a Chief Hardware Officer accused of running the operation from the top. I read the complaint and the reporting around it, and I think the trade-secret claims, real as they might be, are the smaller story here.
By the end of this you’ll have my actual read on what this lawsuit signals about where the AI industry is headed. You’ll also have enough working knowledge of trade secret law to follow the case yourself as it plays out in court over the next year. Not a checklist, more an executive-level take on a fight that’s bigger than the two companies filing it.
What’s alleged: Apple says former employees, including its ex-hardware chief, systematically pulled confidential product information for OpenAI’s hardware push.
What most coverage missed: this lawsuit lands eighteen months after Apple and OpenAI were partners, and six months after Apple quietly picked Google over OpenAI for its own AI.
The legal backdrop most readers don’t know: California bans employee non-competes almost entirely, which is exactly why trade secret law carries so much weight in cases like this one.
My take: a company doesn’t sue a rival it isn’t worried about. This is a confidence problem dressed up as a legal one.
Where this leaves operators: a reminder that vendor alliances in this industry have a shorter shelf life than the contracts built on top of them.
A quick primer: what actually makes something a trade secret, and why this is a California specialty
Before getting into what Apple alleges, it’s worth understanding the legal machinery underneath it, because it explains why this kind of lawsuit is so common in Silicon Valley specifically.
Under both the federal Defend Trade Secrets Act (signed into law in 2016) and California’s own trade secret statute, something only counts as a trade secret if it clears two bars. First, it has to derive real economic value from not being publicly known, meaning a competitor would actually benefit from having it. Second, the owner has to have taken reasonable steps to keep it secret: locked file servers, restricted access, NDAs, exit interviews that revoke credentials. A hardware roadmap, an unreleased prototype’s design specs, or internal manufacturing test procedures all clear that bar easily. A general skill someone picked up on the job, like knowing how to run a good design review, does not.
To win, Apple has to prove three things: a trade secret existed, Apple took reasonable measures to protect it, and it was acquired or used through improper means rather than independent invention. That last piece is where the “LOL” text message and the unreturned laptop actually matter. They aren’t colorful filler for a filing. They’re the evidence Apple needs to clear the “improper means” bar.
Here’s the part that makes California specific. Unlike most states, California voids nearly all employee non-compete agreements outright, under Business and Professions Code Section 16600. A 2024 update went further, making it a separate legal violation for an employer to even include one in a contract. Practically, that means a company like Apple cannot stop a departing engineer from joining a direct competitor the next day, no matter how senior or sensitive their role was. Trade secret law is what’s left. It’s the one legal lever California employers have to stop specific confidential material from walking out the door. They have almost no power to stop the person who holds that knowledge from walking out the door too. That distinction, information versus knowledge, is the entire game these lawsuits are fought over.
This isn’t the first time Silicon Valley has had this exact fight. In 2017, Waymo sued Uber and engineer Anthony Levandowski, alleging he downloaded roughly 14,000 confidential files on self-driving LiDAR technology before leaving to found a startup Uber then acquired. That case settled five days into trial, with Uber handing Waymo about $245 million in equity. Levandowski was later criminally charged, pleaded guilty to one count of trade secret theft in 2020, and was sentenced to 18 months before receiving a pardon in the final hours of the first Trump administration. Apple v. OpenAI is following a similar shape: a departure, a competitor building the same category of product, and a filing that leans hard on internal messages as proof of intent. It’s too early to know if the outcome looks anything like Waymo’s, but the legal template is a familiar one.
What Apple’s complaint alleges
Apple sued OpenAI, its hardware unit io Products, and two former Apple employees, Tang Tan and Chang Liu, in federal court in the Northern District of California, per TechCrunch’s report on the filing. Tan spent 24 years at Apple, most recently as VP of product design for the iPhone and Apple Watch, before leaving to help found io Products with Jony Ive. OpenAI acquired io for $6.5 billion last year and made Tan its Chief Hardware Officer. io is named in the suit. Ive is not.
The most vivid detail belongs to Liu, an eight-year Apple engineer who Apple alleges never returned his company laptop and later found a bug giving him ongoing access to Apple’s internal file servers. He messaged a former colleague: “LOL, I found out I can access the [network storage], so funny,” Bloomberg’s Mark Gurman reported via Fortune. Apple’s filing calls the pattern systemic: “at every level, from members of its Technical Staff to its Chief Hardware Officer… OpenAI has been stealing Apple’s trade secrets and confidential information,” and describes OpenAI’s hardware business as “rotten to its core by its illegal reliance on misappropriated trade secrets.” OpenAI’s response, through a spokesperson: the company has “no interest in other companies’ trade secrets” and is “focused on building innovative technology that empowers people everywhere.” More than 400 former Apple employees have joined OpenAI’s hardware division, per Fortune’s reporting.
That last number matters for reading the case correctly. Apple isn’t alleging that 400 people stole something. Under the legal standard above, hiring hundreds of people away from a competitor is completely legal, and expected in an industry this small. The complaint is narrowly built around specific documents and specific system access tied to two named individuals, which is exactly the kind of surgical claim that survives a motion to dismiss. A broader claim about talent flight alone wouldn’t.
The part I think matters more
Rewind eighteen months. In 2024, Apple and OpenAI announced a partnership putting ChatGPT inside the iPhone. That was the headline at the time: two of the most valuable brands in tech, aligned. Then OpenAI bought Jony Ive’s design studio and started building hardware of its own, explicitly aimed at the category Apple has owned since 2007. Then, in January of this year, Apple quietly signed a $1-billion-a-year deal to run Siri and Apple Intelligence on a custom Google Gemini model instead. ChatGPT became a peripheral option rather than the engine under the hood. Six months after that snub, Apple is in federal court accusing its former partner’s leadership of systematically raiding its talent and its confidential product data.
I don’t think the interesting question is whether Liu downloaded files he shouldn’t have. I think the interesting question is why a company as controlled and press-shy as Apple decided a public lawsuit was the right move here, instead of a quiet settlement or an even quieter internal fix. Companies with real leverage tend to negotiate in private. Companies that feel a genuine platform threat go public, because the lawsuit itself is a signal, to investors, to talent, and to OpenAI, that Apple is willing to fight in the open. Sam Altman has said openly that OpenAI wants to build the device that replaces the smartphone. Apple has spent two decades making sure nothing replaces the smartphone. That’s the actual fight. The trade secrets are the evidence, not the motive.
Where I actually stand on this
I’ll say the quiet part. I’m Team Apple here. Not blindly, the company’s a little past its peak right now, but it’s still a great company through and through. I’ve read both the Steve Jobs biography and Jony Ive’s. The thing that comes through in both is how obsessive Apple has always been about its own design language, doing things Apple’s way. It refuses to ship something that doesn’t feel like Apple, even when that costs time or money. That obsession is the actual moat, not the chips, not the retail stores, the taste.
Say the allegations hold up. Say trade secrets really did make their way into OpenAI’s hardware roadmap with help from people who used to sit in Apple’s own design reviews. I don’t think that’s close. That’s not competition. That’s copying the homework with the answer key someone walked out the door with, and Apple has every right to make an example out of it in open court.
I also don’t think that cancels out the deterrence angle I raised above, and I don’t see that as a knock against Apple for having it. Wanting to slow down a well-funded rival that’s actively poaching your design team is a completely valid business reason to litigate. I’d bet it plays a healthy role in the decision to sue right alongside the actual IP concern. Both things can be true at once. Apple can be genuinely wronged and strategically motivated to make this loud.
The honest tradeoffs
This is one side’s complaint, filed by the party with the most to lose from OpenAI succeeding in hardware. OpenAI denies wrongdoing, and none of the named individuals have responded publicly as of this writing. Nothing here is proven in court, and complaints are written to tell the most damning possible version of events.
Talent moving between competitors is normal, legal, and how this industry has always worked, and California’s ban on non-competes exists specifically to protect that mobility. Apple’s complaint is careful to draw the line around specific documents and system access, not the plain fact that hundreds of its former employees now work at OpenAI. Courts have not been shy about dismissing trade secret claims that amount to “our former employee is now good at their job somewhere else.” If Apple’s evidence turns out to be thinner than the filing suggests, this case could go the same way.
My read on the motive, the confidence-problem framing, is an interpretation, not a fact Apple has confirmed. Apple’s public statement sticks to protecting its IP. I’m connecting dots they haven’t officially connected, and my own Team Apple lean means I’m reading the ambiguous parts of this story in Apple’s favor. A reasonable person could read the same eighteen-month timeline and conclude Apple is using a legitimate legal claim to slow down a competitor it fumbled the partnership with, rather than the other way around.
Where I land
Strip away the AI hardware angle and this is a familiar story. A partnership that made sense when both companies needed something from each other stopped making sense the moment their roadmaps started pointing at the same prize. Call it the Alliance Half-Life, the shrinking window between “strategic partner” and “named defendant” once two companies realize they’re actually building toward the same platform. Eighteen months, in this case. I’d expect that number to keep shrinking as more AI labs decide the real prize isn’t the model, it’s the device people carry.
If you’re building any part of your business on top of a single AI vendor’s platform, a plugin ecosystem, an API, a partnership that felt stable last year, this is worth sitting with. Not because you’ll get sued. Because alliances in this industry are proving to be a lot more temporary than the roadmaps built on top of them assume.
I’ll be watching what OpenAI actually files in response, and whether this becomes the template other AI labs and Big Tech incumbents start reaching for as their roadmaps collide over the next few years. If it is the template, this is the first of many.
If you want to talk through how dependent your own stack is on any single AI vendor’s continued goodwill, that’s a real conversation worth having on an AI Clarity Call, grab a slot at muddventures.com/book.
Reply and tell me if you’re Team Apple on this one too, or if you think I’m cutting them too much slack.
Andrew
P.S. If the theme of keeping control over what you don’t fully own resonates, that’s the same thread running through the Approval Leash post from last week, just applied to a different kind of dependency. Come hang out in the Abra AI community if you want more of this kind of breakdown. And if this is the first issue of mine you’ve read, the archive lives at muddventures.substack.com.