OpenAI says Apple’s own security practices undermine its trade secrets case

We have all seen the headlines about Apple suing former employees or rivals for supposedly walking off with the crown jewels. It is the standard Silicon Valley playbook: hire the best, and if they leave for a competitor, send the lawyers to ensure they do not take a single line of code with them. But the latest clash between OpenAI and Apple has taken a turn that should make every founder and CTO take a hard look at their own offboarding checklists.

New court filings show that OpenAI is not just playing defense; they are going after Apple's own internal hygiene. The core of their argument is simple: If Apple did not care enough to follow basic security protocols when these employees left, how can they claim the information was a strictly guarded trade secret? It is a bold move that highlights a massive gap between corporate policy and actual practice.

The iCloud Loophole

According to the exhibits, a specific incident involving an Apple manager and a former engineer's iCloud account is at the center of the dispute. OpenAI alleges that Apple allowed internal managers to access the personal or semi-personal data of departed staff in a way that contradicts their own privacy and security narratives. In the legal world, if you want to protect a trade secret, you have to prove you took "reasonable measures" to keep it a secret. If you left the door unlocked and the lights on, the court might decide it was not really a secret to begin with.

For builders, this is the nightmare scenario. You spend years developing a proprietary algorithm or a unique dataset, but because your HR team or IT department missed a step during a high-pressure exit, that intellectual property effectively becomes public domain in the eyes of the law. OpenAI is betting that Apple's institutional sloppiness is enough to void their claims.

The Founder's Perspective on Loyalty

In the AI space, the talent war is more like a meat grinder. Engineers move between Google, Anthropic, OpenAI, and Apple every few months. The "trade secrets" being fought over are often just the collective knowledge inside a human brain. When a company like Apple sues, they are trying to put a fence around that brain. But OpenAI's counter-argument suggests that Apple's fence had a massive hole in it.

I have seen this happen in smaller startups too. A founder gets angry that a lead dev left for a rival. They want to sue. But when you look at the logs, the dev still has access to the GitHub repo three weeks after their last day. Or worse, the company allowed them to use a personal Dropbox for work files. You cannot claim something is a million-dollar secret while treating it like a grocery list.

Why Reasonable Measures Matter

The legal standard of "reasonable measures" is intentionally vague, which is why this case is so dangerous for incumbents. OpenAI is pointing out that Apple’s offboarding process was inconsistent. If an Apple manager can just hop into a former employee’s cloud account without a strict, audited protocol, it suggests that the data inside was not being treated with the reverence Apple claims in court.

• Inconsistent Policy Enforcement: If you only enforce security when you are mad at someone, you are not protecting secrets; you are retaliating.
• Digital Sovereignty: The line between personal cloud storage and corporate data is blurrier than ever. If you let employees mix the two, you lose the right to complain when they take their "personal" files with them.
• Managerial Overreach: Giving mid-level managers keys to the kingdom without oversight is a liability, not a feature.

The Skeptical Take

Let’s be real: OpenAI is not doing this because they are champions of worker privacy. They are doing this because it is a highly effective way to shut down a lawsuit that could cost them billions in damages or, worse, result in an injunction against their models. It is a cynical, brilliant legal maneuver. They are effectively saying, "Apple, you are too disorganized to even know what your secrets are, so you shouldn't be allowed to sue us for having them."

This should be a wake-up call for anyone building in the AI or crypto space. We focus so much on the tech—the weights of the model, the security of the smart contract—that we forget the boring administrative stuff. Your legal protection is only as strong as your worst manager's password habits.

What Builders Should Do Now

If you are running a team, do not wait for a lawsuit to audit your offboarding. You need to ensure that the moment a contract ends, the access ends. No "grace periods" for finishing up documentation. No allowing personal cloud backups of sensitive repos. If it is a trade secret, treat it like one from day one to day zero.

OpenAI’s strategy suggests that the best defense against a trade secret heist isn't just a non-disclosure agreement; it's a paper trail of rigorous, unbending security protocols. Apple is a trillion-dollar company and they supposedly tripped over their own shoelaces here. If they can't get it right, your five-person startup definitely isn't getting it right.

The biggest threat to your company's IP isn't a hacker or a competitor; it's the manager who forgot to revoke a former employee's access.

We are entering an era where the "reasonable measures" of the past—a simple NDA and a firm handshake—are no longer enough. The courts are looking for technical enforcement. If you can't prove you blocked the exit, you can't complain that someone walked through it. OpenAI is teaching Apple a very expensive lesson in digital hygiene, and we should all be taking notes.

Takeaway: Your intellectual property is only legally protected if your internal security practices are as rigorous as your legal threats. Sloppy offboarding can turn your most valuable secrets into unprotected data.

---

Read the original at TechCrunch AI →