Apple Says More Former Employees May Have Taken Confidential Data to OpenAI

Apple is broadening the scope of its internal investigation into alleged trade secret transfers, now suggesting that more ex-employees than originally identified may have brought confidential data to OpenAI when they departed.

The pattern here matters more than the headcount. Apple's AI and machine learning teams have lost talent to OpenAI at a steady clip, and the core legal question is whether those transitions carried proprietary technical material — model architectures, training methodologies, or internal tooling — across the boundary.

For engineers watching the AI talent market, this signals something structural. The line between an engineer's internalized expertise and their former employer's trade secrets is genuinely blurry in ML work. What counts as confidential — a novel training approach, a dataset curation strategy, an evaluation framework — is often embedded in how a person thinks rather than in files they can be observed copying. That ambiguity is exactly what makes these cases hard to litigate and harder to prevent.

For OpenAI, the practical exposure depends on what Apple can demonstrate was actually transferred and used. Awareness of Apple's internal direction is not the same as actionable trade secret misappropriation. Courts have consistently required specificity.

For founders and technical leads building AI products, the downstream signal is worth noting: the legal infrastructure around AI talent mobility is tightening. Non-disclosure agreements are being tested in earnest, and companies are investing in departure audits. If you are building a team by hiring from large labs, the due diligence burden around what candidates can and cannot bring with them is real and growing.

This is not resolved. Apple widening the alleged scope suggests the initial investigation surfaced enough to warrant deeper review rather than a clean closure.

Source