Apple has filed a lawsuit against OpenAI, alleging that former Apple employees who later joined the AI company stole Apple’s trade secrets and used them “for the benefit of OpenAI.” The case, which names both individuals and a hardware-related entity tied to Jony Ive, is the latest sign that the competition between major technology companies is no longer confined to software, models, or cloud infrastructure. It is increasingly about who controls the underlying know-how—especially when talent moves across corporate boundaries and when AI firms begin building their own hardware ambitions.
At the center of Apple’s complaint is a claim of pattern-level misconduct rather than a single incident. Apple says it has uncovered “a pattern of theft of Apple’s trade secrets” by OpenAI employees who were previously at Apple. In other words, the lawsuit is framed not as an isolated dispute over one project or one document, but as a broader allegation that confidential knowledge was taken repeatedly and then leveraged after the employees transitioned to OpenAI.
The defendants named in the filing include IO Products, Tang Tan, and Chang Liu. IO Products is described in the complaint as Jony Ive’s hardware startup, which OpenAI reportedly acquired in 2025. Tang Tan is identified as OpenAI’s chief hardware officer, while Chang Liu is alleged to have joined OpenAI from Apple in January. Apple’s theory appears to connect the movement of specific people—particularly those with hardware expertise—to the transfer of sensitive information that Apple believes should have remained protected.
This is not the first time Apple has pursued trade-secret claims, and it is not the first time the tech industry has wrestled with the legal and ethical implications of employee mobility. But the particular combination here—Apple versus OpenAI, hardware leadership, and a company associated with Ive—makes the lawsuit feel like a turning point. It suggests Apple believes the stakes are high enough that it must challenge not only what OpenAI is doing now, but how OpenAI got there.
Why this lawsuit matters goes beyond the immediate parties. Trade-secret litigation often turns on details that the public rarely sees: what exactly was allegedly taken, how it was accessed, whether it was copied or merely referenced, and whether the information was truly confidential under the law. Yet even before those details are fully litigated, the mere act of suing signals Apple’s view that the alleged conduct is serious and potentially systemic.
The hardware angle is especially important. For years, AI companies have been primarily associated with software—training models, deploying APIs, and building user-facing applications. But the industry’s momentum is shifting. OpenAI’s acquisition of IO Products, and its appointment of a chief hardware officer, indicate a strategic push into physical products. When companies move from “AI as a service” to “AI as a device,” the competitive landscape changes. Hardware development involves long timelines, specialized supply-chain relationships, manufacturing constraints, and design decisions that can be deeply proprietary. It also involves engineering processes that are difficult to replicate without internal context.
That context is precisely what trade-secret laws are designed to protect. A trade secret is not just any piece of information; it is information that derives economic value from not being generally known and that is subject to reasonable efforts to maintain secrecy. Apple’s complaint implies that the information at issue meets that standard—and that it was taken by people who had access to it while working at Apple.
There is also a subtle but meaningful narrative choice in Apple’s framing. By emphasizing a “pattern of theft,” Apple is effectively asking the court to treat the allegations as more than a coincidence of overlapping expertise. In many disputes involving employee transitions, the defense often argues that employees simply brought general skills and experience—something that is usually lawful. Apple’s approach suggests it wants to draw a line between general know-how and specific confidential materials or processes.
That distinction will likely become the battleground of the case. Courts typically scrutinize whether the plaintiff can show that the defendant possessed or used specific confidential information, rather than merely benefiting from the employee’s background. Apple’s lawsuit, by naming multiple defendants and tying them to hardware roles, appears intended to strengthen the inference that confidential information was not only accessed but also applied in ways that benefited OpenAI.
The named individuals add another layer. Tang Tan, as OpenAI’s chief hardware officer, occupies a position where decisions about product direction, engineering priorities, and technical strategy converge. If Apple alleges that trade secrets were taken and then used for OpenAI’s hardware efforts, it makes sense that it would target someone in that role. Similarly, Chang Liu’s alleged transition from Apple to OpenAI in January places him in the timeline Apple is likely using to connect access at Apple with subsequent activity at OpenAI.
IO Products, meanwhile, represents the corporate vehicle through which hardware work could be commercialized. Even if the alleged theft involved individuals, Apple’s inclusion of IO Products suggests it believes the company itself may have benefited from the alleged misuse of confidential information. In trade-secret cases, plaintiffs often pursue both the people and the organizations to ensure that any potential remedy can reach the entities that might profit from the disputed technology.
Apple’s public posture so far has been cautious, as is typical in active litigation. A spokesperson statement shared with 9to5Mac begins with a generic assertion about Apple teams developing breakthrough technologies. While that quote does not provide new factual detail, it reflects a common strategy: avoid commenting on specifics that could prejudice the case, while still signaling that Apple views the allegations as credible and worth pursuing.
For OpenAI, the response is not included in the material provided here, but the company’s likely defense path is fairly predictable in trade-secret disputes. OpenAI may argue that any information used was either not confidential, was independently developed, or was general knowledge and skills that employees are allowed to bring with them. It may also challenge the “pattern” characterization, arguing that Apple’s evidence does not establish repeated theft rather than normal overlap in engineering practices.
The court will also likely focus on process evidence: access logs, internal documentation, communications, and any artifacts that Apple claims demonstrate copying or misuse. Trade-secret cases often hinge on whether the plaintiff can show a concrete link between the alleged stolen information and the defendant’s later outputs. That link can be direct—such as identical or near-identical documents—or indirect, such as evidence that confidential information was used in a way that could not be explained by public sources or ordinary engineering reasoning.
This is where the lawsuit becomes more than a headline. The public may see “stolen trade secrets,” but the legal system demands specificity. Apple will need to identify the trade secrets with enough clarity to allow the court to evaluate them. It will also need to show that OpenAI’s actions were connected to those secrets, not merely to the general competence of former Apple employees.
There is also a broader industry question embedded in the case: how should companies handle the tension between protecting confidentiality and enabling legitimate talent mobility? Employee non-disclosure agreements, garden-leave policies, and internal compliance programs are all designed to reduce risk. But in practice, the line between “protected” and “permissible” can blur, especially when employees move into similar technical domains.
In hardware, that blur can be sharper because many engineering problems have convergent solutions. Two teams might arrive at similar designs for reasons unrelated to theft—constraints, physics, and manufacturing realities can force similar outcomes. Defendants in trade-secret cases often lean on this argument: that similarity is inevitable and does not prove wrongdoing. Plaintiffs, conversely, try to show that the similarity is too precise, too contextual, or too dependent on non-public information to be explained away.
Apple’s decision to name a hardware-focused leadership figure and a hardware startup suggests it believes the alleged information is not abstract. It likely relates to tangible design approaches, engineering methods, or internal product development processes that Apple considers uniquely valuable.
Another unique aspect of this lawsuit is the timing and the corporate structure. OpenAI’s reported acquisition of IO Products in 2025 indicates that the hardware effort is not merely speculative—it is organized and resourced. If Apple believes trade secrets were taken and then used, the acquisition could matter in two ways. First, it could provide a pathway for Apple’s claims to reach a company that is actively building hardware. Second, it could complicate defenses about who controlled what information and when. Corporate acquisitions can create messy questions about integration, access, and responsibility.
From a business perspective, the lawsuit also raises questions about how AI companies should manage hardware talent. If Apple’s allegations are accurate, it implies that the risk is not theoretical. If they are not, it still highlights that Apple is willing to use legal tools to deter perceived misuse of confidential information. Either way, the case may influence how future hires are onboarded, how teams are compartmentalized, and how companies document the boundaries between prior employers’ confidential materials and new internal work.
For readers watching the AI-to-hardware transition, the case offers a cautionary lens. The race to build devices powered by AI is not only a race for model performance or user experience. It is also a race for engineering credibility, supply-chain readiness, and product design maturity. Those capabilities are often built through years of internal iteration—exactly the kind of institutional knowledge that companies treat as trade secrets.
If Apple succeeds, the remedies could include injunctions (orders to stop certain activities), damages, or other relief depending on what the court finds. If Apple fails, the outcome could still shape the industry by clarifying what courts consider protectable in the context of employee mobility and hardware development. Either result will likely be studied by legal teams across Silicon Valley and beyond.
There is also a reputational dimension. Trade-secret lawsuits can become public narratives about corporate culture and ethics, even when the facts are contested. Apple’s framing of a “pattern of theft” is designed to communicate seriousness. OpenAI, if it disputes the allegations, will likely emphasize compliance, independent development, and the legality of hiring experienced engineers. The public will watch not only the legal arguments but also how each company positions itself in the broader story of innovation.
What makes this case particularly compelling is that it sits at the intersection of