Apple vs OpenAI: A Trade-Secret Lawsuit That Exposes the Fierceness of the AI Talent War

Apple thinks one of its former engineers took confidential chip-design documents to OpenAI and then tried to cover his tracks. The lawsuit, originally filed on July 10, 2026 in California federal court and dramatically expanded on August 31, 2026, names former senior engineer Chang Liu and his current employer, OpenAI. It accuses Liu of stealing a confidential Apple circuit schematic and an internal engineering tool (sharing a name with an actual Apple product), keeping access through an unreported authentication loophole even after he left, and recruiting an OpenAI colleague to delete files once he learned Apple was investigating him.

It is the sharpest legal manifestation yet of a broader industry pattern: more than 400 former Apple employees now work at OpenAI, and the dispute raises uncomfortable questions about whether the companies vying to build the most capable AI systems are also racing to absorb each other's technical knowledge.

What Apple is actually alleging

According to Apple's court filing, Chang Liu was a senior engineer at Apple who previously worked on the company's neural-engine architecture, the custom silicon Apple designs to power Siri, photo processing, and on-device AI tasks. Before leaving Apple to join OpenAI, Apple says Liu transferred confidential technical material to a personal cloud account.

Once at OpenAI, Apple alleges Liu used two specific pieces of Apple intellectual property: a confidential circuit schematic and an engineering tool that shares a name with an actual Apple internal application. Apple says the overlap between the tool at OpenAI and Apple's proprietary software is too close to be coincidental.

More troubling, Apple claims Liu continued accessing Apple systems after he resigned by exploiting what the filing describes as a "rare, previously unknown authentication bug," a loophole Apple says it had never encountered before. That means, on Apple's account, Liu retained entry to sensitive files long after his formal departure.

The evidence Apple points to includes text conversations that contained "crying laughing" emojis, which Apple argues demonstrate Liu's awareness that he still had access to confidential Apple files even after leaving. And when Liu learned Apple was investigating him, Apple alleges he recruited Yu-Ting Peng, a colleague at OpenAI, to help delete or destroy evidence in June 2026.

What OpenAI says in its defense

OpenAI has pushed back hard. In its own public statements and blog posts, the company argues that Liu only accessed Apple-related data after he formally departed to assist former colleagues, not before. The central argument is that Apple itself failed to promptly revoke system access when Liu left, a common problem across the industry when offboarding is slow or incomplete.

OpenAI denied that any of Apple's trade secrets were used in its work, calling the lawsuit "baseless". The company's position is essentially that lingering access after a resignation is an operational failure on the departing employer's side, not a theft by the employee, and that nothing Apple characterizes as proprietary was actually used at OpenAI.

OpenAI's framing also situates the case within a wider pattern: it has publicly noted the sheer volume of former Apple talent now at OpenAI, suggesting that talent migration in the AI era is so extensive that some overlap in skills and experience is inevitable and not evidence of misappropriation.

Why this case matters for the AI industry

The Apple: OpenAI dispute is not just a legal row between two large companies. It touches several structural features of the current AI race:

• Talent concentration. Apple says more than 400 former employees now work at OpenAI. That number alone signals how aggressively AI labs are poaching from companies with deep hardware and systems expertise, exactly the kind of know-how that makes Apple competitive in custom silicon and on-device ML.
• The hardware-to-AI pipeline. Much of Apple's proprietary technology sits at the intersection of custom silicon and AI inference. If a former engineer can move both the chip designs and the AI-model optimizations to a rival, the competitive implications extend beyond any single product line.
• Evidence preservation. The allegation that Liu asked a colleague to destroy files once he knew he was under investigation moves this from a straightforward IP dispute into potential spoliation territory, and that escalates the legal stakes considerably.
• Authentication gaps. Apple's claim that Liu exploited an unknown authentication bug raises questions about how many legacy accounts across the tech industry remain active after employees leave, a cybersecurity concern that extends well beyond this single case.

What to watch next

Apple is seeking a preliminary injunction that would prevent OpenAI from developing any hardware based on the alleged stolen Apple technology while the case proceeds. It is also asking for expedited discovery, meaning the court would fast-track the evidence-gathering phase so Apple can search for additional proof before anything else is potentially destroyed.

Neither side has gone to trial yet. The case will move through standard civil litigation: disclosure, motions, possible settlement discussions, and, if those fail, a trial on the merits. Should Apple prevail, damages could be substantial; should OpenAI win, the ruling could set a boundary around what counts as protectable trade secret versus general skill and experience in the AI era.

For anyone tracking the broader AI industry, the Apple: OpenAI lawsuit is a canary. It shows that as hardware competence becomes more valuable to AI companies, the legal lines between competition and misappropriation are getting tested in ways they weren't five years ago.

Frequently asked questions

Related articles

Comments