OpenAI Fires Back at Apple Trade Secret Claims With Chat Logs and Emails

A Public Defense in a High-Stakes IP Battle

OpenAI has mounted an unusually public and pointed defense against Apple's trade secret lawsuit, posting chat logs and email threads on its website to counter allegations that it hired Apple engineers to gain access to confidential hardware information. The response, filed in early August, describes Apple's complaint as "careless, aggressive and oddly personal" and argues that the iPhone maker fundamentally misunderstands what happened.

At DailyTechWire, we've tracked rising legal friction between established hardware giants and AI upstarts over talent and IP. This case stands out not just for the parties involved but for the unusually detailed documentary evidence OpenAI chose to make public. The move signals that the company sees the lawsuit as much a reputational battle as a legal one.

Apple filed suit in July against OpenAI and two former employees, Chang Liu and Tang Yew Tan. Liu spent years working on hardware projects at Apple before joining OpenAI; Tan served 24 years at Apple, rising to vice president of product design. Apple alleges that Liu downloaded dozens of confidential files about unreleased products after his departure and that Tan instructed job candidates still employed at Apple to bring physical parts to interviews, using those sessions to extract proprietary information.

The Chat Logs: Who Reached Out to Whom?

OpenAI's central factual rebuttal centers on Liu. The company published communications showing that former colleagues at Apple contacted Liu after he had already left, asking him to help locate certain confidential documents. OpenAI contends that Liu did not seek out this information; rather, Apple employees reached out because they couldn't find files internally and believed Liu might still have residual system access.

According to OpenAI, Apple acknowledged after filing its complaint that its own employees had initiated contact with Liu. Yet instead of accepting responsibility, Apple attempted to "shift the blame to 'residual access,'" framing it as Liu's fault for retaining credentials. OpenAI argues that residual access is a known, widespread problem at Apple caused by inadequate offboarding procedures. Former employees can inadvertently retain access to Apple's internal systems for weeks or months after departure, sometimes without realizing it.

This dynamic is not unique to Apple. Across the Asia-Pacific tech sector, we've observed that rapid headcount growth and distributed IT infrastructure often create gaps in access revocation, particularly when employees move between hardware and software divisions or transfer internationally. What makes this case notable is that the access-control failure has become a central element of litigation rather than an internal audit matter.

The distinction OpenAI draws is significant: if Liu's access persisted because Apple failed to revoke it, and if Apple employees themselves requested his help, the narrative shifts from deliberate theft to organizational sloppiness. That doesn't necessarily absolve Liu of legal liability under trade secret statutes, which can hinge on whether someone knew or should have known they were accessing protected information. But it complicates Apple's portrayal of intentional espionage.

The Tan Allegations and the "Actual Parts" Claim

Apple's allegations against Tan are harder for OpenAI to rebut with documentary evidence. The lawsuit claims Tan directed job candidates, who were still Apple employees, to bring physical components to interviews and that he used those sessions to gather additional proprietary details.

OpenAI's defense here is more general: the company states that Tan has consistently instructed his team not to solicit or use confidential information from other firms. OpenAI emphasizes that Tan's long tenure at Apple, spanning more than two decades, means he accumulated deep domain expertise that is legitimately his to carry forward. The line between knowledge gained through experience and knowledge that constitutes a trade secret is notoriously difficult to draw, especially in hardware design where tacit knowledge, relationships with suppliers, and aesthetic judgment all matter.

The "actual parts" allegation, if proven, would be damaging. Physical prototypes and components carry information that cannot be easily reconstructed from public sources: materials, tolerances, assembly methods, thermal characteristics. Bringing such items to a competitor's interview would cross most companies' bright-line rules. OpenAI has not published interview transcripts or candidate communications to directly refute this claim, suggesting either that such records don't exist or that they are less exculpatory than the Liu chat logs.

The Email Mix-Up: A Minor but Telling Detail

OpenAI also disputes Apple's assertion that it tried to reach OpenAI in February to discuss its concerns but received no response. According to OpenAI, Apple sent its inquiry to the wrong individual after confusing two different Asian surnames. The implication is that Apple's legal and compliance teams were careless in a way that undermines the lawsuit's narrative of deliberate stonewalling.

This detail may seem minor, but it fits OpenAI's broader rhetorical strategy: to paint Apple as acting hastily, without proper diligence, and with a tone that is "oddly personal." The phrase "oddly personal" is unusual in a corporate legal filing and suggests OpenAI believes Apple's motivations extend beyond protecting trade secrets to punishing former employees who left for a rival.

What This Means for Talent Mobility in AI Hardware

The lawsuit arrives at a moment when AI companies are racing to build custom silicon and hardware systems. OpenAI has explored partnerships with semiconductor firms and is reportedly evaluating its own chip designs to reduce dependence on Nvidia. Hiring engineers with deep experience in low-power inference chips, thermal management, and miniaturized form factors is essential to that effort. Many of those engineers have worked at Apple, which has led the industry in integrating custom silicon with tightly controlled software stacks.

Apple, for its part, has invested heavily in on-device AI and is positioning future iPhone and Mac generations around local inference capabilities. Losing senior design talent to OpenAI threatens not just specific projects but the institutional knowledge that makes Apple's hardware integration distinctive.

Across Seoul, Taipei, and Shenzhen, we've seen similar friction as AI labs, cloud providers, and traditional OEMs compete for the same pool of engineers. Taiwan's semiconductor talent has become a particular flashpoint, with both U.S. and Chinese firms accused of aggressive recruiting that borders on IP risk. Export controls and technology transfer restrictions add another layer of complexity, especially when engineers move between jurisdictions.

The OpenAI-Apple case will likely hinge on forensic evidence: server logs showing when Liu accessed which files, metadata on downloaded documents, and testimony from interview candidates about what Tan asked them to bring. If Apple can demonstrate that Liu knowingly accessed files after his departure with the intent to use them at OpenAI, or that Tan orchestrated a systematic effort to extract hardware secrets, the lawsuit will proceed. If the evidence shows residual access that was inadvertent and that Apple's own employees initiated, OpenAI's defense becomes much stronger.

The Broader Legal and Strategic Landscape

This case is part of a larger pattern. In the past 18 months, we've tracked at least a dozen trade secret and non-compete disputes involving AI talent in Asia and North America. Google and Microsoft have both faced similar allegations when hiring from competitors; Anthropic and Cohere have been scrutinized over former OpenAI employees; and Chinese AI labs have been accused of recruiting engineers from U.S. firms under circumstances that raise IP concerns.

What makes this dispute particularly visible is that both parties have chosen to litigate it in public as well as in court. OpenAI's decision to post internal communications is a high-risk move. If additional evidence emerges that contradicts the narrative those logs are meant to support, the company will have amplified its own embarrassment. But if the logs hold up, OpenAI will have successfully reframed the story from "AI upstart steals secrets" to "tech giant fails to manage its own systems and blames departing employees."

For engineers considering moves between hardware and AI firms, the case is a reminder that residual access is a legal minefield. Even if a former employer fails to revoke credentials, using them, or even appearing to use them, can trigger litigation. The safest course is to notify the former employer immediately if access persists, document that notification, and avoid any interaction with former colleagues that could be construed as soliciting information.

For companies, the case underscores the importance of rigorous offboarding. Access revocation should be automated, audited, and logged. When former employees retain access due to administrative failures, the company loses both the moral high ground and, potentially, the legal argument that access was unauthorized.

The outcome of this lawsuit will influence how aggressively companies pursue trade secret claims against rivals who hire their talent, and how much documentary evidence courts require to distinguish between legitimate knowledge transfer and misappropriation. In an industry where the most valuable assets walk out the door every evening, those boundaries matter immensely.