Here's What's on Our Plate Today

• 🔒 Apple sues the partner it once trusted with your iPhone

• 🏜️ TSMC doubles down on desert silicon

• 🎬 Nolan's "Trojan horse" AI warning

• 🛡️ Prompt injection turns on the hackers

Let’s dive in. No floaties needed.

The Laboratory

TL;DR

Apple can lock down the labs, but not the people who know how to build what's inside them.

• Partner to plaintiff: On July 10, 2026, Apple sued OpenAI, hardware startup io Products, and two ex-employees over alleged theft of unreleased hardware trade secrets. OpenAI denies any interest in rivals' secrets.

• The leak is talent: More than 400 former Apple staff now work at OpenAI, carrying design expertise no exit procedure can erase from their heads.

• Secrecy as strategy: Apple's premium model runs on codenames and controlled exits, and trade-secret law is the only lever it has once loyal people choose to leave.

• Where machines break: An unreturned laptop, a lingering software bug, and a still-embedded colleague are exactly the seams where a secrecy machine comes apart.

• Stakes: As every major player chases the post-smartphone AI device, the value of Apple's secrets climbs, and so does the incentive to hire whoever holds them.

The secrets Apple couldn't keep

For nearly two decades, Apple has been the most secretive large company in consumer technology, and it built its business on that secrecy. Products move through development under codenames, prototypes are physically secured inside locked labs, and most employees learn the details of a device only when their own work requires it. This discipline is not decoration; it's part of a culture that preserves the surprise of a launch, protects years of engineering investment, and helps sustain the premium prices Apple's designs command. The company controls its hardware, its software, and increasingly the chips inside its products, a strategy known as vertical integration, meaning it designs most of a product's important parts in-house rather than buying them from others. Secrecy is the glue that holds that approach together.

That world had a comfortable place in it for OpenAI. In June 2024, Apple announced that it would weave ChatGPT into its operating systems, handing the maker of the chatbot access to hundreds of millions of iPhones. The feature shipped that December, letting Siri pass difficult questions to ChatGPT once a user agreed. The two companies were partners, and their interests appeared to align. Apple gained a credible answer to the artificial intelligence wave, and OpenAI gained distribution it could never have built on its own.

A partner in court

On July 10, 2026, that partner became an adversary. Apple sued OpenAI, its chief hardware officer Tang Tan, former Apple engineer Chang Liu, and io Products, the hardware startup OpenAI acquired in 2025 for about $6.5 billion. The complaint accuses them of orchestrating a coordinated effort to obtain trade secrets, confidential business information that gives a company a competitive edge, related to Apple's unreleased hardware. Apple is asking the court to order OpenAI to destroy the allegedly misappropriated material and redesign any forthcoming products that incorporate it. OpenAI denies the allegations, saying, "We have no interest in other companies' trade secrets."

The complaint offers an unusually detailed account of how Apple believes its secrecy broke down. According to Apple, Liu, who spent eight years at the company before leaving in early 2026, retained a company laptop and discovered a software flaw that continued to grant him access to Apple's internal file servers after his departure. In a message quoted in the filing, he wrote to a former colleague, "LOL, I found out I can access the [network storage], so funny." Apple alleges that he then downloaded dozens of confidential files related to unreleased products.

Tan, who spent 24 years at Apple and helped lead the design of the iPhone and Apple Watch, is accused of using Apple's institutional knowledge to recruit talent for OpenAI. Apple says he referred to unreleased projects by their internal codenames during interviews with Apple employees, asked candidates to bring hardware components such as batteries and logic boards to "show and tell" sessions, and circulated a checklist explaining how new hires could avoid Apple's exit security procedures. Apple says it raised these concerns in a letter sent to OpenAI in February 2026, but received no response. None of the allegations has been tested in court.

Partners, then rivals

Whether Apple's allegations ultimately hold up in court, the lawsuit is really about something harder to contain than confidential files: expertise. The lawsuit is less a story about two engineers than about where those engineers went. OpenAI has become one of the largest single destinations for Apple's hardware talent, and Apple's complaint states that more than 400 former Apple employees now work there. Many of them, including Tan, arrived through io Products, the startup co-founded by former Apple design chief Jony Ive to build a screenless, voice-first device meant to create a new product category rather than sit inside Apple's. Ive is not named in the suit. The people Apple trained to keep its secrets are now building a product intended to compete with the iPhone, and they carry in their heads the accumulated knowledge that made Apple's hardware what it is.

That migration is the reason the partnership was always fragile. Apple treats its unreleased designs as its most valuable property, and it has spent years and enormous sums developing them. OpenAI is a software company that quickly realized it needed hardware expertise, and the world's deepest pool of that expertise sat at Apple. The incentive to hire from Apple was obvious, and the results were predictable.

The knowledge lives in people

Underneath the specific accusations sits a contradiction Apple cannot easily resolve. Its secrecy depends on information remaining in the minds of people who choose to remain loyal, yet those people are free to leave, and Apple has little legal power to prevent them from joining a competitor. Trade-secret law is effectively the only tool it has, which is why the entire case is built on it rather than on any promise the engineers made not to compete. The complaint tries to draw a line between the general skill an engineer legitimately carries to a new job and the specific confidential material Apple says was taken, but that line is famously hard to police. A person who spent a decade designing Apple hardware does not forget how Apple designs hardware when they walk out the door.

That legal tension reflects a deeper clash between how the two companies compete. Apple runs what amounts to a secrecy machine, a system of codenames, compartmentalized information, and controlled exits designed to keep knowledge from leaking. OpenAI, racing to catch up in a category it did not previously operate in, ran an aggressive hiring operation that drew that knowledge out of engineers one by one. Apple's complaint describes the point where, in its telling, ordinary recruiting crossed into deliberate extraction, and the mechanics it describes, an unreturned laptop, a lingering software bug, a colleague still inside the company, are exactly the seams where a secrecy machine tends to come apart.

What secrecy is worth now

Apple's complaint describes one instance of how a culture of secrecy can break down. The forces that made it possible, however, extend far beyond this case. The number of people who can design a world-class consumer device is small, they are concentrated at a handful of companies, and every one of those companies is now chasing the same prize, a piece of hardware that could define how people use artificial intelligence after the smartphone.

When Apple, OpenAI, and others are all racing toward that same future, the value of any one company's secrets rises, and so does the incentive to acquire them by hiring the people who hold them. Offboarding security, the routine process of cutting off a departing employee's access, becomes a competitive vulnerability rather than an administrative chore. Apple invested decades in a discipline designed for a world where its best people mostly stayed, but that world no longer exists.

Apple spent decades perfecting a system built on control. It controls the chips inside its devices, the software that runs them, and the hardware that surrounds them. Secrecy was the final layer that held the entire model together. The lawsuit suggests that, in the age of AI, even for multi-billion-dollar companies like Apple, the hardest thing to control is no longer technology but the people who know how to build it.

Prompt of the Day

Bite-Sized Brains

• TSMC is betting the desert on AI: The chipmaker sees "strong multi-year" demand for AI silicon and is pouring more into its Arizona fabs to meet it.

• Nolan spots the Trojan horse: The Odyssey director warned that AI is being wheeled into creative work as a gift that quietly hollows it out from the inside.

• Prompt injection is jamming the hackers: The same trick that fools chatbots is now derailing autonomous AI hacking agents, turning their biggest weakness into a defense.

Tuesday Poll

The Toolkit

• Tabnine: AI coding assistant that runs privately on your own stack, so teams never ship code to public models.

• Runway: AI video and image generator used by filmmakers and marketers to produce cinematic content straight from prompts.

• Superhuman: AI-powered email client that drafts replies, summarizes threads, and drags you to inbox zero faster.

Rate This Edition