Apple v. OpenAI: Trade Secrets and the AI Hardware Race

1. Introduction: From Strategic Alliance to Courtroom Confrontation

1.1. The context of the lawsuit: ChatGPT's integration into the Apple ecosystem and OpenAI's pivot to hardware

On July 10, 2026, Apple Inc. filed a lawsuit in the United States District Court for the Northern District of California, San Jose Division, docketed as Case No. 5:26-cv-07078, accusing OpenAI Foundation, OpenAI Group PBC, its hardware subsidiary io Products LLC, and two former Apple employees — Tang Yew Tan (also referred to as Tang Tan) and Chang Liu — of trade secret misappropriation and breach of contract. The 41-page complaint, drafted by the law firm Weil, Gotshal & Manges LLP, brings claims under the Defend Trade Secrets Act of 2016 (DTSA) (18 U.S.C. § 1836 et seq.) against all defendants, as well as breach-of-contract claims against the two individual employee defendants. The filing marks an open rupture in the relationship between two of the most influential technology companies in the world.

The immediate backdrop to the confrontation traces back to 2024, when Apple and OpenAI formed a strategic alliance to integrate ChatGPT into the iPhone's operating system, using the AI model as a complement to Siri to offer users generative capabilities. This collaboration, framed within the Apple Intelligence strategy, allowed users to access ChatGPT's capabilities through Siri and other company services. The alliance reflected Apple's acknowledgment that it had fallen behind in the AI race unleashed by ChatGPT's arrival on the market. In the words of the complaint itself, Apple had "invested hundreds of billions of dollars and decades of effort in developing innovative consumer hardware products such as iPhone, Apple Watch, and MacBook, together with the enterprise capability and infrastructure necessary to bring them to market at scale."

The relationship began to fray, however, when OpenAI undertook a strategic pivot toward developing its own hardware. The company, which had never specified with precision the exact nature of the device it intended to build, described the effort as a search for "a new way to interact with artificial intelligence that goes beyond traditional products and interfaces." This move placed OpenAI in direct competition with Apple, transforming a former partner into a rival. As trade media have observed, the dispute reflects the growing tension in the technology sector over control of the next generation of AI-powered devices.

1.2. The breakdown of the commercial relationship and the role of the io Products acquisition

The turning point in the relationship came when OpenAI acquired io Products, a hardware startup founded in 2024 by Apple's former chief design officer Jony Ive together with other former Apple executives. The acquisition closed in 2025 for an estimated $6.5 billion. Jony Ive, who had served as Apple's head of design until 2019 and played a decisive role in creating the iPhone, iPod, and iPad, took on the oversight of an AI device project for OpenAI. The complaint, however, does not name Ive as a defendant nor attribute any wrongdoing to him, likely because his involvement in the day-to-day recruiting and development operations is not the subject of the allegations.

The acquisition of io Products, together with the hiring of Tang Yew Tan — a former vice president of product design for the iPhone and Apple Watch for 24 years, now OpenAI's head of hardware — consolidated OpenAI's transformation into a direct competitor of Apple in the consumer electronics market. The complaint asserts that more than 400 former Apple employees currently work at OpenAI, though Apple emphasizes that hiring competitors' talent does not authorize the use of information protected by confidentiality agreements or trade secret law. As a legal analysis by Lowenstein Sandler LLP has noted, the case offers "a high-profile example of the trade secret risks that can arise when a company enters an adjacent market by hiring personnel from an established competitor and working with overlapping suppliers and vendors without imposing strict controls and restrictions to prevent trade secret cross-contamination."

Apple, for its part, redirected its strategy and struck an alliance with Google to use its Gemini model in the upcoming update to its virtual assistant Siri, whose launch is planned for fall 2026. This decision reflects the cooling of relations between Apple and OpenAI following the io Products acquisition. The evolution of the relationship between the two companies — from strategic collaboration in 2024 to courtroom confrontation in 2026 — illustrates the volatility of alliances in the technology sector when commercial interests converge toward direct competition.

1.3. Purpose and structure of this article

This article aims to analyze, from a legal and doctrinal perspective, the complaint in Apple Inc. v. Liu et al., 5:26-cv-07078 (N.D. Cal. 2026), examining its factual foundations, the applicable legal framework — in particular, the Defend Trade Secrets Act of 2016 (DTSA) and the California Uniform Trade Secrets Act (CUTSA) — the conduct allegedly constituting misappropriation, the injunctive relief sought, and the procedural outlook of the litigation.

The complaint, currently in its initial filing stage, brings claims for trade secret misappropriation under the DTSA and breach-of-contract claims against the two individual employee defendants. Apple asks the court for a preliminary injunction barring the defendants from using or disclosing its trade secrets, the return of the misappropriated materials, and monetary damages, including exemplary damages, plus costs and attorneys' fees. OpenAI, for its part, has stated through spokesperson Drew Pusateri that it is reviewing the complaint and has "no interest in using anyone's trade secrets," adding that it remains focused on "building innovative technology that empowers people around the world." An Apple spokesperson, meanwhile, told the BBC that the lawsuit is the result of "significant evidence."

The article proceeds as follows: after this introduction, Section 2 lays out the facts alleged in the complaint and the conduct attributed to each defendant. Section 3 analyzes the applicable legal framework, both federal and state. Section 4 examines the trade secret concept as it applies to this litigation. Section 5 studies the conduct allegedly constituting misappropriation. Sections 6, 7, and 8 address, respectively, corporate liability, the injunctive relief sought, and the damages claims. Section 9 examines OpenAI's likely lines of defense. Sections 10 and 11 analyze the case's significance for the technology ecosystem and its procedural outlook, respectively. Finally, Section 12 presents the conclusions.

2. The facts alleged in the complaint: an overview of Apple's allegations

2.1. The defendants: Chang Liu, Tang Yew Tan, OpenAI Foundation, OpenAI Group PBC, and io Products, LLC

The complaint names five defendants, both entities and individuals. First, OpenAI Foundation and OpenAI Group PBC — the organization's commercial arm — are sued as the entities that, according to Apple, orchestrated and benefited from the alleged scheme to misappropriate trade secrets. Second, io Products LLC, the hardware startup founded by Apple's former chief design officer Jony Ive and acquired by OpenAI in 2025 for approximately $6.5 billion, is likewise named as the vehicle through which OpenAI has channeled its push into the hardware market.

Alongside these entities, the complaint identifies two former Apple employees:

• Tang Yew Tan (also referred to as Tang Tan), a former vice president of product design for the iPhone and Apple Watch for 24 years at Apple, now OpenAI's head of hardware. According to his LinkedIn profile, Tan worked on the iPhone for most of his career at Apple.

• Chang Liu, a former senior systems electrical engineer at Apple for eight years, to whom Apple says it entrusted some of its most sensitive product-development efforts before Liu left the company to join OpenAI in early 2026.

The complaint stresses that more than 400 former Apple employees currently work at OpenAI, though Apple clarifies that hiring competitors' talent does not by itself authorize the use of information protected by confidentiality agreements or trade secret law.

2.2. The central allegation: a "pattern of theft" of trade secrets

The complaint's central thesis is that OpenAI has orchestrated a "coordinated pattern of institution-wide misconduct" to misappropriate Apple's trade secrets. In the complaint's own words: "This case is about former Apple employees stealing Apple's trade secrets to benefit OpenAI. Apple brings this action to put a stop to it."

Apple contends that OpenAI's conduct does not stem from isolated employee actions, but from a systematic strategy deployed "at every level" of the organization. The complaint alleges that, as a result of this conduct, OpenAI's hardware business "rests on the shakiest of foundations, rotten to its core by its unlawful reliance on misappropriated trade secrets."

Apple states it launched an investigation after detecting possible compromises of confidential information and having "uncovered a pattern of theft" of its trade secrets by former employees who had moved to OpenAI. Before filing suit, Apple sent OpenAI a letter in February 2026 setting out its concerns about the possible improper flow of confidential information into OpenAI's business. According to the complaint, OpenAI never responded to that communication.

2.3. The conduct attributed to Liu: downloading confidential files and exploiting a security vulnerability

The complaint attributes to Chang Liu a series of particularly serious actions related to accessing and misappropriating Apple's confidential information after leaving the company.

According to the complaint, when Liu left Apple to join OpenAI in January 2026, he took three things with him:

1. A company MacBook that he never returned.

2. A close relationship with an Apple employee who continued to share internal information.

3. Knowledge of an authentication flaw — a security vulnerability in the software — that let him continue accessing Apple's internal file servers.

Apple alleges that Liu, instead of reporting the security flaw, exploited it to download dozens of confidential hardware-related files from Apple, including technical presentations and specifications for unreleased products. The complaint states that after Liu discovered he could still access Apple's network, he texted a colleague the word "LOL".

Apple further alleges that Liu encouraged another Apple engineer who was interviewing at OpenAI to study confidential Apple materials ahead of her interviews and advised her on how to avoid drawing the attention of Apple's security team when copying files. The complaint states that the two moved their conversations to a private messaging app to evade detection.

2.4. The conduct attributed to Tan: using internal information in the recruiting process and coaching candidates

The allegations against Tang Yew Tan center on his use of OpenAI's recruiting process as a mechanism for extracting confidential Apple information.

Apple contends that Tan, even before leaving Apple, held meetings with OpenAI or its representatives and had conversations with suppliers, in addition to emailing himself information about Apple's suppliers. Once at OpenAI, Tan allegedly reshaped the interview process into a competitive-intelligence tool.

The specific accusations against Tan include:

• Asking candidates who still worked at Apple to bring "real pieces" of Apple hardware to interviews as "props" for "show and tell" sessions. The components requested reportedly included batteries, systems-in-package, multilayer motherboards, shielding, and other hardware components.

• Asking candidates to prepare "Technical Deep Dive" presentations about their work at Apple and to bring CAD files, design artifacts, blueprints, and prototypes.

• Using internal Apple project code names during interviews to question candidates about unreleased products.

• Asking candidates about Apple's prototypes and suppliers.

The complaint quotes a candidate who reportedly said, "I didn't even know we could take that stuff out of the office." In another passage, the complaint describes how a former Apple employee took a screenshot and downloaded files from a "highly confidential Apple project," and Tan later asked about that same project during her interview.

2.5. OpenAI's role as instigator and beneficiary of the scheme

Apple attributes to OpenAI, as an organization, an active role in the alleged trade-secret misappropriation scheme. The complaint contends that OpenAI did not merely benefit passively from information flowing in through its employees, but actively instigated and facilitated the conduct described.

Specifically, Apple alleges that:

• OpenAI advised Apple employees on how to circumvent security procedures when leaving the company, using an internal Apple document describing employee-departure security procedures to do so.

• OpenAI used confidential knowledge of Apple's supplier relationships in its efforts to build a competing hardware business. The complaint states that OpenAI used the misappropriated information to contact Apple's manufacturing partners, at one point asking one of them about Apple's technique for metal finishing on its devices.

• OpenAI has hired more than 400 former Apple employees, which Apple views as indicative of a systematic pattern of talent poaching based, at least in part, on the confidential information these individuals possess. The complaint states that "OpenAI's hiring practices suggest that it hires these individuals at least in part because of the confidential knowledge and specialized Apple experience they have and could improperly obtain."

The complaint concludes that all defendants acted "in concert and as one enterprise, exploiting Apple's confidential information to advance OpenAI's efforts to enter the consumer hardware market." Apple contends that because "misconduct is normalized and exemplified by leadership" at OpenAI, its hardware business "rests on the shakiest of foundations." The complaint also states that the allegations presented are only "the tip of the iceberg," and that Apple "lacks visibility into what has been happening behind closed doors at OpenAI."

3. The legal framework of the lawsuit

3.1. The Defend Trade Secrets Act of 2016 (18 U.S.C. § 1836 et seq.): elements of a misappropriation claim

Apple's complaint is grounded, first and foremost, in the Defend Trade Secrets Act of 2016 (DTSA), codified at Title 18 of the United States Code, Section 1836 and following. Enacted on May 11, 2016, this statute created for the first time a federal civil cause of action for trade secret misappropriation, allowing trade secret owners to sue in federal court regardless of diversity jurisdiction or the amount in controversy.

For a DTSA claim to succeed, a plaintiff must establish three essential elements:

1. The existence of a trade secret: the information must be that which "derives independent economic value, actual or potential, from not being generally known to, and not being readily ascertainable through proper means by, another person who can obtain economic value from the disclosure or use of the information." In addition, the owner must have taken "reasonable measures" to keep it secret.

2. Misappropriation of the trade secret by the defendant: the DTSA defines misappropriation as acquiring a trade secret through "improper means" — including theft, bribery, misrepresentation, inducement of a breach of a duty to maintain secrecy, or espionage — or disclosing or using the secret without consent. The concept of "improper means" is analogous to that used in the California Uniform Trade Secrets Act (CUTSA).

3. A connection between the trade secret and a product or service used in, or intended for use in, interstate or foreign commerce: this connection to interstate commerce forms the basis of federal jurisdiction.

To these elements, case law adds the requirement that the plaintiff demonstrate that the defendant acquired, disclosed, or used the trade secret through improper means, including theft, deception, or breach of confidentiality obligations. The DTSA also provides for exemplary damages of up to twice the amount of actual damages awarded when the trade secret is shown to have been misappropriated "willfully and maliciously."

Apple bases its DTSA claim on the allegation that the defendants acquired, disclosed, and used its trade secrets through improper means, in particular through the individual employee defendants' breach of confidentiality duties and OpenAI's inducement of those breaches.

3.2. The California Uniform Trade Secrets Act (Cal. Civ. Code § 3426 et seq.): particularities of the state regime

Alongside the federal claim, the complaint invokes the California Uniform Trade Secrets Act (CUTSA), codified at California Civil Code Sections 3426 through 3426.11. California was one of the first states to adopt the Uniform Trade Secrets Act (UTSA), and its statute has a structure and language similar to the uniform law. CUTSA defines "improper means" in a manner consistent with the DTSA, including "theft, bribery, misrepresentation, breach or inducement of a breach of a duty to maintain secrecy, or espionage through electronic or other means."

CUTSA nonetheless presents some particularities relevant to this litigation:

• Statute of limitations: a misappropriation claim must be brought within three years of when the misappropriation was discovered or, through the exercise of reasonable diligence, should have been discovered.

• Protection of the secret during litigation: the court must preserve the confidentiality of an alleged trade secret through reasonable means.

• Broader scope of protection: California's version of the UTSA affords employers broader protection against employee theft or misappropriation of trade secrets than the original uniform law.

The concurrence of the federal and state claims — DTSA and CUTSA — allows Apple to invoke both regimes simultaneously without contradiction, given the substantial overlap in how each defines the elements of the cause of action.

3.3. The breach-of-contract claims: Apple's Intellectual Property Agreement and its confidentiality provisions

In addition to its trade secret misappropriation claims, Apple brings breach-of-contract claims against the two individual employee defendants, Chang Liu and Tang Yew Tan.

Apple alleges that both employees signed, upon joining the company, an Intellectual Property Agreement imposing confidentiality obligations both during and after their employment. The complaint contends that Liu and Tan breached those obligations by:

• Misappropriating Apple's confidential information and using it to benefit OpenAI.

• Failing to return Apple's confidential materials after leaving.

• Disclosing Apple's confidential information to third parties.

• Breaching the duties of loyalty and confidentiality inherent to their status as trusted employees.

These contractual claims are independent of the trade secret misappropriation claims and seek, in addition to monetary damages, specific performance of the contractual obligations — in particular, the return of all of Apple's confidential materials and information in the defendants' possession.

3.4. The concurrence of federal and state claims: procedural issues and choice of forum

Apple's complaint raises the concurrence of federal and state claims in a single proceeding, which is permissible under the U.S. legal system so long as the two bodies of law do not conflict and the court has jurisdiction over all of them.

The choice of the Northern District of California — sitting in San Jose — as the forum is no accident. This district, which covers the heart of Silicon Valley, houses the United States District Court for the Northern District of California and has vast experience with trade secret and intellectual property litigation in the technology sector. The case's assignment to Magistrate Judge Virginia K. DeMarchi suggests that the case will initially proceed under this court's procedural rules, known for their efficiency in managing complex litigation.

The concurrence of DTSA and CUTSA claims in the same proceeding nonetheless raises some procedural questions:

• Nature of the claims: the DTSA claim is federal in nature and may be heard by federal courts regardless of the amount in controversy, whereas the CUTSA claim is a state-law claim that could be heard by state courts.

• Choice of forum: Apple has chosen the federal forum, bringing the DTSA claim and, alongside it, the CUTSA claim under supplemental jurisdiction or diversity jurisdiction.

• Burden of proof: under both regimes, the burden of proof rests on the plaintiff, who must establish the existence of the trade secret, the adoption of reasonable measures to protect it, and the defendant's misappropriation.

Apple's strategy of combining federal and state claims aims to maximize its chances of success and broaden the range of available remedies, since the DTSA allows for exemplary damages and CUTSA provides for specific injunctive remedies, even though the two regimes are substantially aligned.

As Lowenstein Sandler LLP's legal analysis has noted, the case offers "a high-profile example of the trade secret risks that can arise when a company enters an adjacent market by hiring personnel from an established competitor and working with overlapping suppliers and vendors."

4. The trade secret concept in the litigation: categories of protected information

4.1. The legal definition of a trade secret under the DTSA and CUTSA

Whether the information Apple alleges was stolen actually qualifies as a "trade secret" as a matter of law is the logical and legal predicate for the entire misappropriation claim. Both the Defend Trade Secrets Act of 2016 (DTSA) and the California Uniform Trade Secrets Act (CUTSA) contain substantially overlapping definitions of this concept, though with relevant nuances.

The DTSA, in Section 1839(3), defines a "trade secret" as:

all forms and types of financial, business, scientific, technical, economic, or engineering information, including patterns, plans, compilations, program devices, formulas, designs, prototypes, methods, techniques, processes, procedures, programs, or codes, whether tangible or intangible, and whether or how stored, compiled, or memorialized physically, electronically, graphically, photographically, or in writing, if: (A) the owner thereof has taken reasonable measures to keep such information secret; and (B) the information derives independent economic value, actual or potential, from not being generally known to, and not being readily ascertainable through proper means by, another person who can obtain economic value from the disclosure or use of the information.

CUTSA's definition, found at Section 3426.1(d) of the California Civil Code, is analogous, though phrased somewhat differently:

"trade secret" means information, including a formula, pattern, compilation, program, device, method, technique, or process, that: derives independent economic value, actual or potential, from not being generally known to the public or to other persons who can obtain economic value from its disclosure or use; and is the subject of efforts that are reasonable under the circumstances to maintain its secrecy.

Both definitions therefore require the concurrence of three cumulative elements:

1. The secret nature of the information: the information must not be generally known or readily ascertainable by third parties.

2. The adoption of reasonable measures by the owner to preserve confidentiality.

3. The independent economic value derived from the secrecy.

Case law has clarified that "independent economic value" can be actual — if the information already generates benefits — or potential — if its disclosure would harm the owner or benefit a competitor. The DTSA also expressly recognizes that a compilation of public information can constitute a trade secret when combined or compiled in a novel way, even if each individual element is generally known.

4.2. The categories of information alleged by Apple: technical specifications, designs, product roadmaps, manufacturing processes, and supplier relationships

Apple's 41-page complaint identifies a broad range of information categories that, according to the company, constitute trade secrets and were allegedly misappropriated by the defendants. The complaint describes the information allegedly stolen as "the crown jewels of Apple's hardware machine."

The categories of information alleged by Apple include, according to the complaint and published analyses:

a) Hardware engineering information: circuit and system design specifications, component architecture, and power-management innovations. The complaint specifically mentions information on power and battery systems.

b) Product designs and blueprints: CAD (Computer-Aided Design) files, design artifacts, blueprints, and physical prototypes of unreleased products. The complaint alleges that the defendants asked candidates to bring these materials to interviews.

c) Product roadmaps: milestone timelines, development schedules, and product plans not disclosed to the public.

d) Manufacturing and finishing processes: proprietary metal-finishing techniques and processes, as well as information on manufacturing design and process engineering.

e) Supply-chain information: confidential details about suppliers and contractors within Apple's supply chain, including the internal terminology Apple uses in its communications with suppliers.

f) Specialized engineering data: information on coexistence and electromagnetic interference, as well as AI and machine-learning integration data developed for Apple's hardware products.

g) Testing and development methods: testing methodologies and component-development processes.

h) Business relationships: information about supplier relationships and other business relationships.

The complaint specifies that the stolen information includes both physical materials — components, prototypes, devices — and digital information — CAD files, presentations, internal documents. It also mentions the use of internal Apple project code names to extract information during interviews.

Apple emphasizes that the allegedly stolen information is not limited to currently marketed products, but extends to unreleased products and future developments. This circumstance is particularly relevant to the assessment of damages, since premature disclosure of information about unannounced products can deprive Apple of the competitive advantage derived from the "surprise factor" of its launches.

4.3. The reasonable measures Apple adopted to preserve confidentiality: the secure repository system, the "need to know" policy, and confidentiality agreements

The complaint devotes significant attention to describing the "reasonable measures" Apple has adopted to preserve the confidentiality of its information — an essential requirement under both the DTSA and CUTSA. Apple must show it took measures proportionate to the nature and value of the information to keep it secret.

According to the complaint and published analyses, Apple has implemented a trade secret protection system that includes the following elements:

a) Confidentiality and intellectual property agreements: every Apple employee signs, upon joining the company, an Intellectual Property Agreement (IPA) whose obligations survive termination of employment. This agreement details the employees' confidentiality and non-disclosure obligations, including a prohibition on disclosing proprietary information without Apple's written consent. The IPA also prohibits employees from "taking any document, material, or copy thereof, whether on paper or any other medium, containing proprietary information." Employees also commit to following Apple's information security rules, to using their best efforts to safeguard confidential information, and to not allowing third parties to make such disclosures.

b) Annual business-conduct training: employees receive annual "Business Conduct" training that reiterates confidentiality principles.

c) "Need to know" policy: even within Apple, confidential information should only be shared on a need-to-know basis. Apple's internal documents are marked "Need to Know" to limit their circulation.

d) Secure network file repository: Apple uses a secure network file repository with database access-permission policies.

e) Third-party confidentiality agreements: Apple requires confidentiality agreements from contractors and other third parties who access its confidential information.

f) Physical and technical security controls: the complaint describes access-security controls for Apple's facilities and computer systems.

These measures, according to Apple, demonstrate that the company made "reasonable efforts" to keep its information secret, thereby satisfying the second requirement of the legal definition of a trade secret.

4.4. The dispute over the secret nature of the information: relevant case-law criteria

Whether the information alleged by Apple actually constitutes a trade secret will likely be one of the central battlegrounds of the litigation. OpenAI, in its defense, may dispute that the information at issue was truly secret or that Apple took reasonable measures to protect it.

Courts have established several criteria for determining whether information is secret, including:

• Public knowledge: information that is generally known or accessible to the public does not constitute a trade secret.

• Ease of ascertainment: information that can readily be determined through other legitimate means — such as reverse engineering — is not a trade secret.

• Protective measures: the existence of protective measures — such as those described above — is a relevant, though not determinative, indicator of secrecy.

• Economic value: the information must have economic value derived from its secrecy, meaning that its disclosure would harm the owner or provide an advantage to competitors.

• Duration of secrecy: how long the information has been kept secret is also a relevant factor.

In this case, Apple will need to show that information about manufacturing processes, unreleased product designs, and supplier relationships meets these criteria. The company contends that the allegedly stolen information is not generally known, is not readily ascertainable, and has been subject to rigorous protective measures.

OpenAI's defense, however, could argue, among other lines:

• That some information — such as metal-finishing techniques — is known within the industry or can be reverse-engineered.

• That Apple did not adopt sufficient protective measures to prevent Liu from accessing its systems after he left.

• That information about supplier relationships is not a trade secret but common industry knowledge.

5. Analysis of the conduct allegedly constituting misappropriation

5.1. Acquisition of trade secrets by improper means: "show and tell" during job interviews

Apple's complaint frames the first category of conduct constituting misappropriation — acquisition of trade secrets through "improper means" — around a systematic practice of asking job candidates who still worked at Apple to bring confidential information and materials to interviews at OpenAI.

The complaint describes in considerable detail the instructions Tang Yew Tan, OpenAI's head of hardware and former Apple employee, allegedly gave to candidates. According to the complaint, Tan asked applicants to prepare presentations called "Technical Deep Dive" on their work at Apple, including CAD files, design artifacts, blueprints, and prototypes, and to bring "real pieces" of Apple hardware to interviews such as batteries, systems-in-package, multilayer motherboards, shielding, and other hardware components. The complaint quotes a candidate who reportedly said, "I didn't even know we could take that stuff out of the office."

In another passage, the complaint describes how Tan allegedly used internal Apple project code names during interviews to question candidates about unreleased products. Apple describes these practices as a form of "industrial espionage" carried out through improper means, since they induce candidates — still bound by confidentiality duties — to reveal information to which the interviewer would not otherwise have legitimate access.

Trade secret case law has held that actively soliciting confidential information from a competitor's employees constitutes an "improper means" when the person soliciting it knows or should know that the information is protected by confidentiality agreements. In this sense, the conduct attributed to Tan, and by extension OpenAI, would fit within the concept of inducing a breach of a confidentiality duty, a scenario expressly contemplated in the definition of "improper means" under both the DTSA and CUTSA.

5.2. Disclosure or use of trade secrets without consent: the alleged use of Apple information in OpenAI's hardware development

The second form of misappropriation alleged by Apple consists of the disclosure and use of the stolen trade secrets for the development of OpenAI's hardware business. Apple contends that OpenAI not only acquired confidential information through improper means, but has actively used it to build its hardware infrastructure, contact suppliers, and advance the development of its own AI device.

The complaint alleges that OpenAI used the stolen information to (i) contact Apple's manufacturing partners, (ii) ask one of them about Apple's technique for metal finishing on its devices, and (iii) use supplier-relationship information to position itself within the supply chain. These allegations point to active use of the trade secrets — not merely their passive possession or disclosure — an essential element of a misappropriation claim.

The complaint states that "OpenAI, which had never before built any hardware at all, has been competing effortlessly in hardware development" during the period in which it allegedly stole Apple's information. Apple contends that the speed and sophistication of OpenAI's hardware development are indicative of a reliance on stolen information, and characterizes OpenAI's hardware business as "rotten to its core by its unlawful reliance on misappropriated trade secrets."

Under the DTSA, a plaintiff need not prove that the defendant made direct use of the trade secret; it suffices to show that the secret was used to accelerate or improve the development of a competing product. Case law has established that use of a trade secret can be inferred from the nature of the product developed or from the speed with which it was developed, even absent direct proof of the use of the stolen information.

5.3. Inducement to breach confidentiality duties: OpenAI's role in advising Apple employees to evade security controls

The complaint attributes to OpenAI an active role in inducing breaches of confidentiality duties, a specific form of "improper means" under both statutes. Apple contends that OpenAI advised Apple employees on how to circumvent security procedures when leaving the company, using an internal Apple document describing employee-departure security procedures to do so.

Inducement to breach confidentiality duties can take various forms, including explicit instruction to breach the duty, rewarding the disclosure of confidential information, or creating an organizational climate in which breaching confidentiality is tolerated or encouraged. In this case, the complaint alleges that OpenAI created precisely this kind of climate by normalizing misappropriation, as shown by the fact that Tan coached candidates on how to avoid detection by Apple's security team.

OpenAI's liability for inducing breaches of confidentiality duties also extends to the conduct of its executives and leadership, who, according to Apple, "normalized and exemplified" the misconduct at every level of the organization. This allegation is relevant to vicarious liability and to the eventual award of exemplary damages, since the DTSA requires, for such damages to be awarded, that the misappropriation have been "willful and malicious" (18 U.S.C. § 1836(b)(3)(C)).

5.4. The employees' liability for breach of their contractual and fiduciary obligations

In addition to the trade secret misappropriation claims, the complaint brings breach-of-contract claims against the two individual employee defendants, Chang Liu and Tang Yew Tan. Apple contends that both breached the confidentiality obligations imposed by the Intellectual Property Agreement they signed upon joining the company.

Liu's liability stems from his conduct after leaving Apple: downloading confidential files through the security flaw he exploited, failing to return his company MacBook, and advising a candidate on how to avoid detection when copying files. Tan's liability stems from his use of the interview process as a competitive-intelligence tool, his requests for confidential materials from candidates, and his advising Apple employees on how to circumvent security procedures.

The complaint alleges that Liu and Tan, by breaching their post-employment confidentiality obligations, engaged in breach of contract and, additionally, in breach of fiduciary duty owed to Apple as trusted employees. An employee's fiduciary duty to their employer includes a duty of loyalty, which prohibits the employee from using the employer's confidential information for their own benefit or that of a third party, even after employment ends, in certain circumstances.

Liu and Tan's personal liability is independent of OpenAI's liability and could carry significant financial consequences for them, including the return of the stolen materials, monetary damages — potentially exemplary — and costs and attorneys' fees. The concurrence of employee liability and corporate liability also raises the question of joint and several liability, which the court will need to examine.

As a Business Insider analysis has observed, one of the goals of Apple's lawsuit is to force open OpenAI's "black box," since Apple "had no other way to determine what information OpenAI had taken." The discovery phase could reveal the true scope of the alleged misappropriation and determine whether it extends to additional categories of trade secrets not specified in the initial complaint.

6. Corporate liability: OpenAI and io Products

6.1. The doctrine of vicarious liability (respondeat superior) and its application to the case

Apple's complaint does not limit itself to attributing conduct to the two individual employee defendants, Chang Liu and Tang Yew Tan, but extends liability to the corporate defendants OpenAI Foundation, OpenAI Group PBC, and io Products, LLC (the "Corporate Defendants"). The legal basis for this extension of liability rests on the doctrine of respondeat superior — an employer's vicarious liability for the acts of its employees — as well as on the allegation that the Corporate Defendants instigated, facilitated, and benefited from the misappropriation conduct.

Courts in the Northern District of California and other federal courts have repeatedly recognized that a corporate entity can be held liable for trade secret misappropriation committed by its employees under the respondeat superior doctrine. As the case law cited in Extreme Reach, Inc. v. Spotgenie Partners, LLC notes, "courts in this district recognize that a corporate defendant 'may be held liable for [an individual employee's] misappropriation of trade secrets under the doctrine of respondeat superior.'" Other California courts have reached the same conclusion, holding that "the employer is vicariously liable when the employee's trade secret misappropriation 'was committed within the scope of employment.'"

The key element for applying the respondeat superior doctrine is that the employee's conduct occurred "within the scope of employment." In this case, Apple alleges that Liu and Tan's conduct — downloading confidential files, using the interview process to extract information, and advising other employees on how to evade security controls — occurred precisely in the exercise of their duties at OpenAI. Tan, as OpenAI's head of hardware, was acting within his responsibilities in designing the recruiting process and developing the company's hardware strategy. Liu, as an OpenAI engineer, accessed and used the stolen information in the context of his work for the company.

The respondeat superior doctrine is particularly relevant to this litigation because it allows Apple to bring claims directly against OpenAI and io Products for their employees' conduct, without needing to prove that OpenAI's leadership specifically knew of or authorized each act of misappropriation. As courts have noted, vicarious liability applies even when "others at [the company] did not know of these actions, because the corporate defendant was liable under respondeat superior." This doctrine applies under both CUTSA and the DTSA.

Apple's complaint, however, goes further than vicarious liability and alleges that the Corporate Defendants actively instigated and facilitated the misappropriation conduct, making them directly liable and not merely vicariously so.

6.2. The allegation of a "coordinated pattern of institution-wide misconduct"

Apple's complaint makes a particularly serious allegation: that the misappropriation conduct does not stem from isolated acts by rogue employees, but from a "coordinated pattern of institution-wide misconduct" orchestrated by OpenAI itself. If proven, this allegation would establish OpenAI's direct — not merely vicarious — liability for the misappropriation, and could support an award of exemplary damages under the DTSA (18 U.S.C. § 1836(b)(3)(C)).

Apple contends that the misconduct was "normalized and exemplified by leadership" at OpenAI. This assertion rests on the following factual allegations:

1. Tan, as an OpenAI executive, reshaped the interview process into a competitive-intelligence tool, coaching candidates to bring confidential Apple materials. This conduct, which Apple describes as a "strategy to extract confidential Apple information," would not have been possible without the acquiescence or support of OpenAI's leadership.

2. OpenAI advised Apple employees on how to circumvent security procedures, using an internal Apple document describing the company's departure procedures to do so. This allegation suggests institutional knowledge of the information-extraction strategy.

3. OpenAI hired more than 400 former Apple employees, which Apple views as indicative of a systematic pattern of talent poaching based, at least in part, on the confidential information these individuals possess. Apple states that "OpenAI's hiring practices suggest that it hires these individuals at least in part because of the confidential knowledge and specialized Apple experience they have and could improperly obtain."

4. OpenAI used information about Apple's supplier relationships to contact its manufacturing partners, at one point asking one of them about Apple's technique for metal finishing on its devices. This allegation points to institutional, not merely individual, use of the stolen information.

5. Apple sent OpenAI a letter in February 2026 raising its concerns, and OpenAI never responded. Apple presents this lack of response as evidence that OpenAI had no intention of ceasing its conduct.

The complaint concludes that all defendants acted "in concert and as one enterprise, exploiting Apple's confidential information to advance OpenAI's efforts to enter the consumer hardware market." Characterizing the conduct as a "coordinated pattern" is relevant not only to establishing OpenAI's direct liability, but also to proving the "willful and malicious" nature of the misappropriation, a requirement for exemplary damages under the DTSA.

6.3. io Products' position as OpenAI's hardware vehicle and its involvement in the alleged scheme

The complaint names io Products, LLC as a defendant — the hardware company founded in 2024 by Apple's former chief design officer Jony Ive together with other former Apple executives, and acquired by OpenAI in 2025 for an estimated $6.5 billion. Naming io Products as a defendant reflects Apple's thesis that the alleged trade secret misappropriation occurred in the context of, and for the benefit of, OpenAI's hardware business, of which io Products is the primary vehicle.

The complaint does not name Jony Ive as a defendant nor attribute any wrongdoing to him. This omission, according to published analyses, likely reflects practical considerations — Ive was not involved in day-to-day recruiting, engineering, or development operations — as well as Ive's personal relationship with Laurene Powell Jobs, Steve Jobs's widow, who still carries weight at Apple.

io Products' liability is framed in the complaint around several points:

1. io Products is the vehicle through which OpenAI has channeled its push into the hardware market. The acquisition of io Products, together with the hiring of Tan as OpenAI's head of hardware, consolidated OpenAI's transformation into a direct competitor of Apple in the consumer electronics market. Apple alleges that the stolen information was used precisely to accelerate io Products' hardware development.

2. io Products directly benefited from the stolen information. Apple contends that OpenAI and io Products used confidential knowledge of Apple's supplier relationships in their efforts to build a competing hardware business. The complaint states that OpenAI used the stolen information to contact Apple's manufacturing partners and to benefit from proprietary manufacturing techniques.

3. io Products participated in the recruiting process that, according to Apple, was the primary mechanism for extracting information. While the complaint does not detail io Products' specific involvement in interviews, its inclusion as a defendant reflects the thesis that OpenAI's entire hardware operation — including io Products — benefited from the stolen information.

io Products' liability is grounded both in the respondeat superior doctrine — for the acts of its employees and executives — and in its alleged direct participation in the "coordinated pattern" of misappropriation. Apple asks that any injunctive relief and damages award extend to io Products, preventing it from using or benefiting from Apple's trade secrets.

7. The injunctive relief Apple seeks

7.1. The request for a preliminary injunction: barring use and disclosure of trade secrets

Apple's complaint includes a request for a preliminary injunction as one of the primary remedies sought from the court. Apple asks the District Court for the Northern District of California to issue an order barring all defendants — OpenAI Foundation, OpenAI Group PBC, io Products LLC, Chang Liu, and Tang Yew Tan — from possessing, using, or disclosing its trade secrets and confidential information. The request for injunctive relief is grounded in the urgency of the situation: Apple contends that the misappropriation conduct is ongoing, and that absent immediate measures, the harm to its competitive interests will be irreparable.

Specifically, the preliminary-injunction request seeks to:

1. Bar the defendants from possessing, using, or disclosing any Apple trade secret or confidential information.

2. Order the return of all Apple materials, documents, components, prototypes, and equipment in the defendants' possession.

3. Order the destruction or deletion of any copy, reproduction, or summary of such information, including information stored in computer systems.

4. Order the preservation of evidence to ensure that documentation relevant to the litigation is not destroyed or altered.

Apple has stated it will move "promptly" to formally request the preliminary injunction from the court. The immediate risk for OpenAI is that, if granted, the injunction could materially delay or disrupt its consumer hardware program — a project in which the company has invested roughly $6.5 billion. As industry analysis has noted, an injunction blocking the use of Apple-related information in hardware development could delay product timelines and raise doubts about the company's valuation.

7.2. The request for a permanent injunction: return of materials and destruction of information

Alongside the preliminary injunction, Apple seeks a permanent injunction that, if granted following resolution of the litigation on the merits, would permanently bar the defendants from possessing, using, or disclosing Apple's trade secrets. A permanent injunction would constitute the final remedy and would have lasting effect, unlike a preliminary injunction, which operates on an interim basis while the case proceeds.

The permanent injunction request pursues the same goals as the preliminary measure, but on a final rather than merely provisional basis:

1. Permanent bar on possessing, using, or disclosing trade secrets.

2. Final return of all Apple materials and equipment.

3. Destruction or deletion of every copy or reproduction of confidential information.

Granting a permanent injunction would require the court to find, after trial, that the defendants engaged in trade secret misappropriation and that a permanent cease-and-desist order is warranted. A permanent injunction would have an even more severe impact on OpenAI's hardware business than the preliminary measure, since it would indefinitely bar the company from using any information the court determines constitutes an Apple trade secret.

7.3. Requirements for granting injunctive relief in trade secret matters: likelihood of success, irreparable harm, and balance of interests

Granting a preliminary injunction under U.S. law is subject to strict requirements, which courts have systematized in what is known as the "Winter test" (Winter v. Natural Resources Defense Council, 555 U.S. 7, 2008). For a court to grant preliminary injunctive relief, the plaintiff must establish the following elements:

1. Likelihood of success on the merits: the plaintiff must show it is likely to prevail on the merits of the case. In the DTSA context, this requires showing the existence of a trade secret, the adoption of reasonable measures to protect it, and the defendant's misappropriation.

2. Irreparable harm: the plaintiff must show that, absent injunctive relief, it will suffer harm that cannot be adequately remedied through monetary damages. In trade secret matters, unauthorized disclosure or use of confidential information is generally considered irreparable harm, since the loss of exclusivity over the secret cannot be precisely quantified.

3. Balance of equities: the court must weigh the plaintiff's interests — avoiding irreparable harm — against the defendant's interests — bearing the restriction on its activity — and determine that the balance tips in the plaintiff's favor.

4. Public interest: the court must consider whether granting the injunction serves the public interest, which in trade secret matters is identified with protecting intellectual property and fair competition.

The burden of proof for obtaining a preliminary injunction rests entirely on the plaintiff, Apple. The company will need to file a motion for preliminary injunction supported by evidence — declarations, documents, expert reports — establishing all four requirements.

Lowenstein Sandler LLP's legal analysis has noted that if granted, Apple's request for injunctive relief could "materially delay or disrupt" OpenAI's hardware program. Even a narrower injunction — requiring OpenAI to quarantine certain documents, restrict employee access, or cease using specific manufacturing processes or suppliers — could significantly affect the development of its device. As TechCrunch has observed, "even setting aside whether the court grants any kind of injunction or restraining order on what OpenAI is doing, this can naturally lead to that kind of situation."

The court could also opt for a partial or modified injunction, not barring all of OpenAI's hardware activity, but only the conduct deemed most directly tied to the misappropriation. In any case, the mere filing of the preliminary-injunction request and the commencement of discovery could have a chilling effect on OpenAI's activity, generating uncertainty about the viability of its hardware project.

8. The damages claims

8.1. Compensatory damages: actual losses and OpenAI's unjust enrichment

Apple's complaint seeks compensatory damages to make the company whole for the economic losses it suffered as a result of the alleged misappropriation of its trade secrets. These damages are grounded in two distinct, though complementary, legal concepts: Apple's actual losses and OpenAI's unjust enrichment.

First, Apple seeks damages for the actual losses the misappropriation caused it. According to the complaint, these losses include:

1. Investigation and mitigation costs: the expenses Apple incurred investigating the alleged misappropriation, including the internal investigation that led to the discovery of the "pattern of theft" and efforts to mitigate the resulting harm.

2. Loss of competitive advantage: the economic harm resulting from OpenAI's ability to accelerate its entry into the hardware market by using Apple's confidential information, thereby avoiding the costs and time normally required to develop such capability from scratch. The complaint alleges that the stolen information "has been used to advance OpenAI's hardware development to Apple's detriment."

3. Harm to reputation and business relationships: the harm caused to Apple's relationships with its suppliers and business partners as a result of the alleged misuse of information about its supply chain.

Second, Apple seeks restitution of the unjust enrichment OpenAI obtained as a result of the misappropriation. Under this theory, the plaintiff does not necessarily seek an amount equivalent to its own losses, but rather the amount by which the defendant was unjustly enriched at the plaintiff's expense. Here, Apple contends that OpenAI has economically benefited from the stolen information by avoiding the costs of developing its own hardware and by accelerating its entry into the market.

The exact amount of compensatory damages Apple seeks is not specified in the complaint, and it will fall to the trial phase to determine, absent a settlement. Lowenstein Sandler LLP's legal analysis notes that Apple could seek damages "in an amount equal to the actual damages suffered," and could pursue an unjust-enrichment theory to claim the profits OpenAI obtained as a result of the misappropriation.

8.2. Exemplary or punitive damages: the allegation of intentional and malicious misconduct

Alongside compensatory damages, Apple's complaint seeks exemplary or punitive damages against the defendants. The DTSA, in Section 1836(b)(3)(C), provides for exemplary damages "in an amount not exceeding twice the amount of actual damages awarded" when the trade secret is shown to have been misappropriated "willfully and maliciously." CUTSA, in Section 3426.3(c), contains a similar provision, authorizing an award of exemplary damages "not exceeding twice the actual damages" when misappropriation occurred "willfully and maliciously."

To support an award of exemplary damages, Apple must establish the concurrence of the following elements:

1. Willfulness of the misappropriation: the defendant's conduct must have been intentional, not merely negligent.

2. Maliciousness of the misappropriation: the defendant's conduct must have been motivated by a desire to harm the trade secret owner or by conscious disregard for its rights.

Apple alleges that the defendants' conduct meets these requirements. In particular, the complaint contends that:

• Misconduct was "normalized and exemplified by leadership" at OpenAI, demonstrating a deliberate and systematic pattern.

• The defendants acted in a coordinated manner to extract Apple's confidential information, evidencing a deliberate intent to misappropriate the trade secrets.

• Apple sent OpenAI a letter in February 2026 raising its concerns, and OpenAI never responded, which Apple presents as evidence of bad faith.

• Tan and Liu acted with knowledge, instructing others on how to evade security procedures and exploiting security vulnerabilities.

An award of exemplary damages could have a significant economic impact on OpenAI, since the DTSA and CUTSA allow for doubling the amount of actual damages awarded. If Apple were to establish substantial actual damages before the court, the addition of exemplary damages could raise the total award considerably.

8.3. Costs and attorneys' fees

In addition to compensatory and exemplary damages, Apple's complaint seeks costs and attorneys' fees from the defendants. The DTSA, in Section 1836(b)(3)(D), provides that a court may award the plaintiff "reasonable attorneys' fees" when the misappropriation is shown to have occurred "willfully and maliciously." CUTSA contains a similar provision at Section 3426.4, authorizing a court to award "reasonable attorneys' fees" to the prevailing party when misappropriation occurred "willfully and maliciously."

An award of costs and fees is therefore an ancillary claim contingent on a prior finding that the defendants' conduct was willful and malicious. If the court finds that the misappropriation was willful and malicious, Apple would be entitled to recover the litigation expenses it incurred, including the fees of its legal representation — Weil, Gotshal & Manges LLP — which in litigation of this magnitude can run into the millions of dollars.

Apple asks the court to include the following in any final award:

1. Costs of suit, including court filing fees and service-of-process expenses.

2. Attorneys' fees Apple incurs in litigating the case.

3. Any other relief the court deems just and proper.

Though ancillary, the claim for costs and fees can have an additional deterrent effect on the defendants, increasing the economic cost of the litigation if the court finds their conduct was willful and malicious. As legal analysis has noted, the combination of compensatory damages, exemplary damages, and costs could result in a total award substantially higher than mere compensation for actual losses.

9. OpenAI's defense: likely arguments and strategies

9.1. OpenAI's denial: "we have no interest in other companies' trade secrets"

From the very day Apple filed the complaint, OpenAI has articulated a public response in two distinct phases. Its first reaction, issued hours after the complaint was filed on July 10, 2026, was a brief statement asserting that the company had "no interest in using anyone's trade secrets" and remained focused on "building innovative technology that empowers people around the world." This initial statement, provided to TechCrunch, did not address the merits of the allegations and adopted a tone of institutional distancing.

OpenAI's response entered a second phase on July 14, 2026, when the company issued a more substantive statement asserting: "While we take these allegations seriously, we are not aware of any evidence supporting the merits of this lawsuit." This statement was shared publicly by Bloomberg journalist Ed Ludlow and picked up by TechCrunch and other outlets. In the same statement, OpenAI added: "We believe in fair competition and in people's freedom to work where they choose, and we're focused on building innovative technology that empowers people around the world."

OpenAI spokesperson Drew Pusateri had previously told the BBC and AP News that the company "has no interest in using anyone's trade secrets" and remains focused on "building innovative technology that empowers people around the world." This position, however, does not constitute a formal legal response to the complaint, but rather an institutional statement foreshadowing the defensive line OpenAI may pursue in the proceeding.

The timing OpenAI chose for its second statement — four days after the complaint was filed — suggests a calculated strategy: avoiding an impulsive reaction that could be used against it in the proceeding, while sending a message of confidence to investors and the market. As Bloomberg has noted, this statement came while OpenAI continues to work on a screenless mobile device described as a "humanized AI companion" that would compete directly with Apple's products.

9.2. Possible lines of defense: absence of a secret, absence of misappropriation, absence of reasonable measures, and lack of causation

OpenAI's legal defense in Apple Inc. v. Liu et al., 5:26-cv-07078, could be built around several lines of argument, each attacking an essential element of the trade secret misappropriation claim. Published legal analyses identify the following possible defensive strategies:

a) Absence of a trade secret

OpenAI could dispute that the information Apple alleges actually constitutes a "trade secret" within the meaning of the DTSA and CUTSA. To do so, the defense could argue that:

• Information on manufacturing techniques — such as metal finishing — is known within the industry or can be legitimately reverse-engineered.

• Information about suppliers and business relationships is not secret, but rather common knowledge within the sector.

• Information about unreleased products, even if not public, does not meet the "independent economic value" requirement derived from secrecy.

The burden of proving the secret nature of the information rests on Apple, which will need to show that the information is not generally known or readily ascertainable by other means. As attorney Patricia Lantzy told Business Insider, Apple must prove that "the information at issue was truly secret, that Apple took precautions to safeguard it, and that the defendants intentionally obtained it through improper means."

b) Absence of misappropriation

OpenAI could argue that no "misappropriation" occurred in the legal sense, because:

• The conduct of the individual employee defendants does not constitute "improper means" under the DTSA and CUTSA.

• The information Liu and Tan had access to was not "acquired" through improper means, but was part of the general knowledge and experience they gained during their employment at Apple.

• There is no evidence that OpenAI has "used" the trade secrets in developing its hardware, beyond Apple's bare allegations.

As to Liu, OpenAI could argue that his downloading of files — if it occurred — does not constitute misappropriation if the information was not truly secret or if Liu was authorized to access it at the time of the download. As to Tan, OpenAI could contend that the interview process was not designed to extract confidential information, but to assess candidates' technical competence.

c) Absence of reasonable measures

OpenAI could dispute that Apple adopted "reasonable measures" to preserve the confidentiality of its information, an essential requirement of the trade secret definition. The defense could argue that:

• Apple did not implement sufficient security controls to prevent Liu from accessing its systems after he left, as shown by the security flaw Liu allegedly exploited.

• Apple did not adequately supervise its employee-departure process, allowing Liu to keep a company MacBook and Tan to circulate information about departure procedures.

• Apple's security measures were not proportionate to the value of the information they were meant to protect.

d) Lack of causation and unquantifiable damages

Finally, OpenAI could argue that there is no direct causal link between the alleged misappropriation and the damages Apple claims. The defense could contend that:

• OpenAI's hardware development has not benefited from Apple's information, but is the product of the experience and talent of its own engineers.

• The damages Apple alleges are speculative and cannot be quantified with certainty.

• OpenAI's entry into the hardware market has not caused Apple harm that can be remedied through damages.

e) A favorable precedent: xAI v. OpenAI

OpenAI could also invoke a recent judicial precedent in its favor. In June 2026, U.S. District Judge Rita Lin dismissed a similar trade secret misappropriation lawsuit brought by xAI, Elon Musk's AI company, against OpenAI. The court found that xAI had failed to prove that OpenAI pressured former engineer Xuechen Li to share proprietary details about its Grok model. Although the underlying facts differ — the xAI case involved pressuring an employee to share information about AI models, not the theft of hardware secrets — the precedent could be invoked by OpenAI to argue that Apple's allegations are equally speculative.

However, as legal analyses have observed, Apple's case presents substantial differences: Apple's complaint is supported by concrete factual allegations — file downloads, retention of devices, instructions to candidates — that go beyond mere suspicion. The trial court has not yet ruled on the merits, and OpenAI has not, at this point, filed a formal motion to dismiss or a detailed response to Apple's allegations.

9.3. The distinctive character of California's labor market: employee mobility and the inevitable disclosure doctrine

One of the most distinctive aspects of this litigation is that it unfolds within California's legal framework — a state that has deliberately adopted a policy of protecting employee mobility and rejecting non-compete agreements. This particularity of California law decisively shapes OpenAI's possible defenses and the limits of Apple's claims.

The rejection of the "inevitable disclosure" doctrine

California has expressly rejected the "inevitable disclosure" doctrine, which in other jurisdictions allows an employer to prevent a former employee from working for a competitor on the theory that, in the new position, the employee will inevitably disclose or use the former employer's trade secrets. The leading case here is Whyte v. Schlage Lock Co., 101 Cal. App. 4th 1443 (2002), in which the California Court of Appeal held that:

"the inevitable disclosure doctrine ... creates a de facto covenant not to compete ... [and] runs counter to the strong public policy in California favoring employee mobility."

As AI-governance executive Jean Gan noted on LinkedIn, cited by Business Insider: "California courts have largely rejected the inevitable disclosure doctrine, and the state won't enforce non-competes, so Apple can do nothing about the 400 former employees now at OpenAI." Indeed, the mere fact that OpenAI has hired more than 400 former Apple employees does not, by itself, provide a basis for a trade secret misappropriation claim in California.

Apple's strategy: focusing on conduct, not mere hiring

Aware of this limitation, Apple has framed its complaint not around the mere hiring of former employees — which would be legally irrelevant in California — but around specific conduct that, according to Apple, constitutes misappropriation. As Jean Gan has observed:

"So every allegation rests on conduct: retained devices, unauthorized access, misused documents, coached evasion. In a jurisdiction where talent moves freely by design, trade secrets law is the only legal perimeter left around institutional knowledge, and Apple has pleaded squarely inside it."

This strategy is legally sound and reflects a deep understanding of the particularities of California law. Apple is not suing OpenAI for hiring its former employees — which would be untenable in California — but for the specific conduct those former employees and OpenAI allegedly engaged in to obtain and use confidential information. The complaint centers on allegations of active conduct: retaining devices, unauthorized access to systems, downloading documents, coaching candidates to evade security controls, and using supplier information.

The risk of trade-secret "cross-contamination"

Lowenstein Sandler LLP's analysis has noted that Apple's case illustrates the risks that can arise "when a company enters an adjacent market by hiring personnel from an established competitor and working with overlapping suppliers and vendors without imposing strict controls and restrictions to prevent trade secret cross-contamination." This observation is particularly relevant to OpenAI, which has hired more than 400 former Apple employees and established relationships with the same suppliers Apple uses.

OpenAI's defense could argue that it has implemented sufficient controls to prevent "cross-contamination" of trade secrets. However, Apple's allegations — which include coaching candidates to bring physical materials to interviews and using information about Apple's departure procedures — suggest otherwise. As Jean Gan has observed: "Supply chains move trade secrets just as easily as departing staff do, and few confidentiality frameworks treat them with the same rigor." Apple alleges precisely that OpenAI used information about its suppliers to benefit from proprietary manufacturing techniques, including metal finishing.

The burden of proof and the role of discovery

Ultimately, OpenAI's defense faces a significant procedural challenge: the discovery phase could reveal the true scope of the alleged misappropriation. As a Business Insider analysis has noted, Apple filed suit partly because it "had no other way to determine what information OpenAI had taken." Discovery — which could include access to OpenAI's internal documents, emails, and messages — could give Apple the evidence it needs to establish its allegations, or it could exonerate OpenAI by showing that no misappropriation occurred.

OpenAI's position, for now, is one of categorical denial: it is "not aware of any evidence" supporting the lawsuit. This statement, however, is a public statement, not a formal answer to the complaint. The real legal battle will be fought in the courts, through discovery and, eventually, at trial.

10. The case's significance for the technology ecosystem

10.1. The impact on the relationship between AI companies and hardware manufacturers

Apple's lawsuit against OpenAI is not an isolated dispute, but a symptom of a structural shift in the technology industry: the convergence of artificial intelligence and physical hardware. As technology analyst Paolo Pescatore, founder of PP Foresight, has observed, "this case underscores that the next phase of the AI battle will be fought in factories and living rooms, not just data centers." "The AI race is moving beyond models and chatbots to who controls the device, the interface, and the direct relationship with the consumer." "As AI becomes physical, industrial design, components, manufacturing expertise, and supply-chain relationships are becoming just as valuable as algorithms."

This observation places the litigation in its broader context: the contest over "physical AI." Apple and OpenAI compete not only in the market for language models, but in the market for consumer devices that incorporate artificial intelligence — phones, computers, robotics, and wearables. OpenAI has been developing a smartphone with AI agents that could directly compete with Apple's iPhone, and court filings indicate the company is also exploring custom-designed headphones. OpenAI CEO Sam Altman himself has described the forthcoming device as more "peaceful and calm" than an iPhone.

Apple, for its part, had already fallen behind in the AI race: its much-hyped AI-enhanced Siri only recently began rolling out after a two-year delay, and was built on Google's Gemini model rather than its own AI model. Apple chose Google precisely because of OpenAI's hardware ambitions, which it viewed as a threat. By filing suit, Apple is "conceding that it needs outside help in AI to compete and feels threatened by OpenAI, regardless of whether Sam Altman's company actually stole the technology."

The litigation therefore carries a strategic dimension beyond the mere trade secret claim itself. As trade-secrets attorney Jeffrey Farrow, partner and chair of the litigation department at Michael Robinson, has observed: "This lawsuit is not just about the trade secret claim — it is, in my view, an early indication of the AI battle moving from software to physical products." "For the past several years, tech companies have competed over foundation models and cloud infrastructure, and in the future they're going to compete over who builds the devices people interact with every day."

10.2. The tension between technological collaboration and competition in the device market

The evolution of the relationship between Apple and OpenAI — from partners in 2024 to litigants in 2026 — illustrates the fragility of tech alliances when the parties' strategic interests converge toward direct competition. The original collaboration, integrating ChatGPT into the Apple ecosystem, was based on a seemingly clear division of labor: Apple would provide the hardware and distribution; OpenAI, the AI model. However, OpenAI's decision to develop its own hardware — culminating in the $6.5 billion acquisition of io Products — transformed a partner into a direct competitor. Apple, in turn, redirected its Siri AI strategy toward Google, closing the door on future collaboration with OpenAI.

The case reflects a broader tension in the tech ecosystem: collaboration between AI companies and hardware manufacturers carries a structural risk of conflicting interests. AI companies, once they have built model capabilities, tend to expand into hardware to capture value at the device layer and in the direct consumer relationship. Hardware manufacturers, in turn, seek to bring AI capabilities in-house to avoid depending on outside vendors. This dynamic creates a collaboration-competition cycle that trade secret law is called upon to manage.

The litigation also raises questions about the future of strategic alliances in the sector. If AI companies and hardware manufacturers cannot trust that the other party will not use knowledge gained during a collaboration to compete directly, the trend will be toward vertical integration or more restrictive agreements on confidentiality and non-competition. As Yahoo Finance's analysis has noted, "there's so much at stake here that a misstep by OpenAI, which no longer enjoys a stellar reputation, could drag down a significant part of the AI ecosystem with it." "Stock market participants hope OpenAI can survive this without triggering an AI-bubble burst."

10.3. Implications for labor mobility in the tech sector: between free competition and trade secret protection

The Apple v. OpenAI case places at the center of the debate one of Silicon Valley's oldest and most persistent tensions: employee mobility versus trade secret protection. Stanford Law School professor Mark Lemley has observed that Apple's lawsuit "has the potential to be a really important case," but has cautioned that some of Apple's allegations, such as hiring hundreds of former Apple employees, "aren't illegal in California, where Silicon Valley itself grew up in part because of state laws that let employees leave for a competitor." Lemley qualifies: "But if Apple's claims that employees took confidential documents — and that OpenAI is using those documents — are true, that is indeed a problem for OpenAI."

This nuance is crucial. California law rejects non-compete agreements and the inevitable disclosure doctrine. An employee is free to leave a job and work for a competitor, even if in the new role they will inevitably draw on general knowledge and experience gained at their prior employer. What an employee cannot do is take confidential documents, prototypes, or specific information protected by confidentiality agreements. This distinction — between general knowledge and specific trade secrets — is the axis on which the litigation turns.

Apple's complaint reflects a shift in strategy among tech companies for managing talent attrition. Unable to rely on non-compete agreements, Apple has instead chosen to use trade secret law as a mechanism to deter its employees from taking confidential information to a competitor. As a Business Insider analysis has noted, Apple "can't just sue over the 400 departures"; instead, each claim must be grounded in specific conduct: retained devices, unauthorized access, misused documents, coached evasion.

The talent war between Apple and OpenAI is fierce. OpenAI has hired more than 400 former Apple employees, many drawn by "massive and ultra-lucrative" compensation packages. Talent attrition on the iPhone's product-design team became so severe that Apple was forced to partially rebuild its team structure. Apple responded with "unusually high" retention bonuses and even sent executives to persuade senior engineers to stay. Recruiting did not stop once Apple's investigation began: as recently as June 2026, OpenAI hired away Apple's head of smart glasses.

The litigation could have a chilling effect on labor mobility in the tech sector. Former Apple employees working at OpenAI may become more reluctant to discuss their prior work, and executives will avoid certain lines of questioning. The result is an organization "slower and more cautious, exactly when speed is what matters." As Bloomberg has noted, "the allegations and the investigation are enough to hurt OpenAI."

10.4. The echo of Waymo v. Uber: an unsettling precedent for OpenAI

The Apple-OpenAI litigation bears significant parallels to Waymo v. Uber (2017), one of the most famous trade secret cases in Silicon Valley history. In that case, Waymo — Google's self-driving-car division — sued Uber over alleged misappropriation of trade secrets related to autonomous-vehicle technology. Former Waymo engineer Anthony Levandowski had downloaded 14,000 confidential files before joining Uber. The case settled for $245 million and led to Levandowski's departure from Uber.

The similarities to the present case are notable:

1. The theft of files: in both cases, a former employee allegedly downloaded confidential documents before leaving the company. In Apple's case, Liu allegedly downloaded "dozens of confidential hardware-related files."

2. Hiring away a competing team: both Uber and OpenAI hired a significant number of former employees from the plaintiff company. OpenAI has hired more than 400 former Apple employees.

3. Entry into a competing market: both Uber and OpenAI entered a market in direct competition with the plaintiff company. OpenAI is developing hardware that would compete with the iPhone.

4. Allegations of a systematic pattern: in both cases, the plaintiff alleged that the conduct was not isolated, but part of a "coordinated pattern." In Waymo v. Uber, Waymo's spokesperson stated: "This is not a case of an engineer who accidentally took a few files with him. This is a systematic effort to steal Waymo's technology."

5. The risk of an injunction forcing a product redesign: in Waymo v. Uber, the risk that a court would order Uber to redesign its self-driving technology was a key factor in the settlement. In this case, if Apple shows that OpenAI built its secrets into a product, a court could order a redesign. This "would echo Apple's settlement with chip startup Rivos."

The Waymo v. Uber precedent suggests that, even if OpenAI ultimately prevails on the merits, the cost of the litigation — in terms of distraction, uncertainty, and possible injunctive relief — could be significant. As PitchBook has noted, Apple's case "echoes Waymo's lawsuit against Uber a decade ago." "Apple is taking the AI race to court."

10.5. The impact on OpenAI's IPO and the AI bubble

The timing Apple chose to file suit is no accident. OpenAI is at a critical juncture in its corporate trajectory: it has confidentially filed for an IPO and is expected to announce its first hardware product by late 2026, with a launch planned for 2027. The lawsuit threatens to overshadow both milestones.

The lawsuit's immediate impact is operational and psychological. A trade secret lawsuit forces "new legal reviews, tighter internal controls, and hours of depositions and discovery that pull engineers away from building." The result is an organization "slower and more cautious." Bloomberg Intelligence expects Apple to obtain "an early, narrow injunction that quarantines the disputed material and forces OpenAI to preserve evidence and certify compliance." "None of that needs a jury."

The impact on the IPO could be even more significant. Bloomberg reports that "the damage is already arriving: chilled recruiting, a device launch under a cloud, and a defensive posture right as OpenAI heads toward its IPO." "Apple may not need to win to slow down the company most determined to end the iPhone era." Legal uncertainty and the possibility of injunctive relief could cool investor appetite and delay or complicate OpenAI's path to going public.

More broadly, the litigation could test the AI bubble. OpenAI is a central piece of the AI ecosystem and a participant in Project Stargate, the U.S. government's initiative to secure American dominance in artificial intelligence. Oracle, which has a $638 billion backlog, relies on OpenAI for half of it. "There's so much at stake here that a misstep by OpenAI, which no longer enjoys a stellar reputation, could drag down a significant part of the AI ecosystem with it." "Stock market participants hope OpenAI can survive this without triggering an AI-bubble burst."

11. Procedural outlook and future scenarios

11.1. The current state of the proceeding: service of process and judicial assignment

The complaint in Apple Inc. v. Liu et al., 5:26-cv-07078, was filed on July 10, 2026, in the United States District Court for the Northern District of California, San Jose Division. As of this writing — July 20, 2026 — the proceeding is at its initial filing stage. The 41-page complaint has been assigned to Magistrate Judge Virginia K. DeMarchi of the Northern District of California, who handles complex intellectual-property and trade-secret litigation.

The specific procedural status, as of today, is as follows:

1. The complaint has been filed and docketed under case number 5:26-cv-07078.

2. Service of process on the defendants — OpenAI Foundation, OpenAI Group PBC, io Products LLC, Tang Yew Tan, and Chang Liu — is underway or has already been completed. Service is an essential procedural requirement for the court to acquire personal jurisdiction over the defendants.

3. The defendants have not yet filed a formal answer to the complaint. The deadline for filing an answer in the Northern District of California is generally 21 days from service, extendable by agreement of the parties or court order.

4. Apple has stated its intention to move for a preliminary injunction "promptly." This request will be formalized through a motion for preliminary injunction supported by documentary evidence and declarations.

5. There is no indication that OpenAI has yet filed a motion to dismiss under Rule 12(b)(6) of the Federal Rules of Civil Procedure. If OpenAI pursues this route, the proceeding could see significant delay while the court rules on the sufficiency of the allegations.

The assignment of the case to Magistrate Judge DeMarchi is relevant for several reasons. First, magistrate judges in the Northern District of California have experience managing complex intellectual-property and trade-secret litigation and typically apply an efficient procedural schedule. Second, Magistrate Judge DeMarchi may handle discovery matters and incidental motions, though substantive issues and final injunctive relief could be referred to the assigned district judge. Available sources do not specify whether the case has been referred to a district judge or whether Magistrate Judge DeMarchi will handle the merits, but standard practice in the Northern District of California is for preliminary-injunction motions to be decided by the district judge.

11.2. Likely procedural milestones: motion to dismiss, discovery, and trial

Absent a settlement, the litigation will follow a typical procedural path in the Northern District of California. The foreseeable procedural milestones are as follows:

a) Motion to Dismiss

OpenAI, as defendant, could file a motion to dismiss under Rule 12(b)(6) of the Federal Rules of Civil Procedure, arguing that the complaint fails to state "a claim upon which relief can be granted." If granted, this motion could result in dismissal of the complaint, either allowing Apple to amend it (dismissal without prejudice) or barring it from doing so (with prejudice).

Arguments OpenAI could raise in a motion to dismiss include:

• That the complaint fails to identify the allegedly stolen trade secrets with sufficient particularity.

• That Apple's allegations of a "coordinated pattern" are mere conclusory allegations lacking sufficient factual support.

• That the complaint fails to show Apple adopted reasonable measures.

• That the claim is time-barred, though this allegation seems premature given that the allegedly unlawful conduct occurred in recent months.

Filing a motion to dismiss could delay the proceeding by several months while the court rules on the motion and, if warranted, grants Apple leave to amend the complaint.

b) Discovery

The discovery phase is likely the most important — and probably the most costly and prolonged — procedural milestone of the litigation. During discovery, the parties exchange information relevant to the case, including documents, emails, messages, declarations, and testimony. The Northern District of California typically sets a discovery schedule that can run from 6 to 12 months, depending on the case's complexity and the volume of documents involved.

In this litigation, discovery faces particular challenges:

• The confidential nature of the information: Apple will need to disclose some of its trade secrets during discovery to establish their existence and value, creating a risk of further disclosure. To mitigate this risk, the court will issue a protective order limiting access to confidential information to the parties' attorneys and experts.

• The volume of information: Apple and OpenAI generate massive amounts of data, and discovery could involve reviewing millions of documents, emails, and messages.

• The conduct of former employees: testimony and declarations from Liu, Tan, and other former Apple employees will be crucial to determining whether misappropriation occurred.

As a Business Insider analysis has noted, Apple filed suit in part because it "had no other way to determine what information OpenAI had taken." Discovery could provide Apple the evidence it needs to establish its allegations, or it could exonerate OpenAI by showing that no misappropriation occurred.

c) Motion for Summary Judgment

After discovery closes, either party could file a motion for summary judgment under Rule 56 of the Federal Rules of Civil Procedure, asking the court to rule without trial where there is no genuine dispute of material fact and the moving party is entitled to judgment as a matter of law. In trade secret litigation, summary judgment motions are common and typically focus on issues such as the existence of the trade secret, the adoption of reasonable measures, or the absence of misappropriation.

d) Trial

If the case is not resolved through a motion to dismiss or summary judgment, it will proceed to trial, before a jury or the judge, depending on the parties' election. Trial could last several weeks or months, depending on the complexity of the evidence and expert testimony. Available sources do not specify a likely trial date, but complex litigation in the Northern District of California typically takes 18 to 36 months from filing to trial.

11.3. Alternative scenarios: settlement, interlocutory injunctive relief, or trial-court resolution

The Apple-OpenAI litigation could resolve through several alternative scenarios, each carrying different implications for the parties.

a) Settlement

Settlement is a frequent outcome in trade secret litigation, especially when the parties have long-term commercial interests and litigation costs are high. The Waymo v. Uber precedent, resolved through a $245 million settlement, suggests that a settlement is a plausible scenario here. Factors that could favor a settlement include:

• Litigation costs: both sides face attorneys' and experts' fees that could run into the tens of millions of dollars.

• Uncertainty of outcome: Apple's claims are not assured of success, especially under California law.

• The need to preserve business relationships: although the Apple-OpenAI relationship has soured, both companies may have an interest in keeping communication channels open for future collaborations in non-competing areas.

• The interest in avoiding injunctive relief: a settlement could prevent the court from issuing an injunction that would disrupt OpenAI's hardware program.

A settlement is not inevitable, however. Apple's complaint is framed in very harsh terms — describing OpenAI's hardware business as "rotten to its core" — and the reputational harm to OpenAI could be difficult to repair through a monetary settlement. Apple may also be interested in obtaining a judicial ruling that sets an unfavorable precedent for OpenAI and other AI companies attempting to enter the hardware market.

b) Preliminary Injunction

As discussed in Section 7, Apple will seek a preliminary injunction barring the defendants from possessing, using, or disclosing its trade secrets. Whether this injunction is granted or denied will be a crucial procedural milestone, as it will determine the scope of restrictions on OpenAI's activity while the litigation proceeds. If the court grants the injunction, OpenAI could be forced to interrupt or significantly alter its hardware program, potentially devastating its business strategy and its valuation ahead of an IPO. If denied, Apple would lose a key point of leverage and would need to await resolution of the case on the merits.

The court could also opt for a partial or modified injunction, not barring all of OpenAI's hardware activity, but only the conduct deemed most directly tied to the alleged misappropriation.

c) Trial on the Merits

If the case reaches trial and is resolved at the trial-court level, the court will issue a ruling finding that misappropriation did or did not occur, and, if applicable, granting the injunctive relief and damages Apple seeks. The trial-court ruling could be appealed by either party to the U.S. Court of Appeals for the Ninth Circuit, which could extend the litigation for several additional years.

11.4. The Waymo v. Uber precedent and other analogous litigation in the tech sector

The Apple-OpenAI litigation finds its closest parallel in Waymo LLC v. Uber Technologies Inc., 3:17-cv-00939 (N.D. Cal. 2017), which has become a touchstone in Silicon Valley trade secret litigation. A comparative analysis of both cases offers clues for anticipating the procedural outlook of the present litigation.

Waymo v. Uber: case summary

In February 2017, Waymo — Alphabet's self-driving-car division — sued Uber Technologies and its subsidiary Otto for trade secret misappropriation. Former Waymo engineer Anthony Levandowski had downloaded 14,000 confidential files before leaving the company and founding Otto, which Uber later acquired for $680 million. Waymo alleged that Uber had used the stolen trade secrets to accelerate its self-driving-vehicle program.

The case bore the following similarities to the present litigation:

• The theft of a significant number of confidential files (14,000 in Waymo, "dozens" in Apple).

• Hiring away a competing team (Levandowski and other Waymo engineers joined Uber; more than 400 former Apple employees have joined OpenAI).

• Entry into a competing market (Uber entered the self-driving-vehicle market; OpenAI is entering the consumer hardware market).

• Allegations of a "coordinated pattern" and "institution-wide misconduct."

The outcome: settlement and injunction

Waymo v. Uber settled in February 2018, barely a year after it was filed. The settlement terms included:

• A payment of $245 million from Uber to Waymo, equivalent to 0.34% of Uber's equity at the time.

• Uber's commitment not to use Waymo's trade secrets in its self-driving technology.

• Anthony Levandowski's departure from Uber.

The settlement came after District Judge William Alsup issued a partial preliminary injunction in May 2017, requiring Uber to return the stolen documents and submit to a rigorous discovery process. The injunction, though it did not bar Uber from continuing its self-driving program, created significant pressure on the company and accelerated the settlement.

Lessons for the Apple v. OpenAI case

The Waymo v. Uber precedent offers several lessons for the present litigation:

1. Injunctive relief is a decisive point of leverage: in Waymo v. Uber, the preliminary injunction forced Uber to undergo rigorous discovery and return documents, creating pressure that led to the settlement. Apple is likely seeking a similar effect in this litigation.

2. Settlement is the most likely outcome: trade secret litigation between tech companies is often resolved through settlement to avoid the costs and uncertainty of trial.

3. Reputational harm can be as costly as damages: in Waymo v. Uber, the reputational harm to Uber was significant, especially in the context of its 2019 IPO. OpenAI, currently preparing for its own IPO, could be similarly affected.

4. Hiring a competitor's former employees carries risk, but is not illegal per se: Professor Mark Lemley has noted that, as in Waymo v. Uber, the mere hiring of a competitor's former employees is not illegal in California, but the theft of documents and use of confidential information is. The distinction between general knowledge and specific trade secrets will be key in both cases.

Other relevant litigation

Beyond Waymo v. Uber, this case fits within a tradition of trade secret litigation in the tech sector that includes:

• Google v. Uber (2019): Google sued Uber over alleged misappropriation of trade secrets related to self-driving-vehicle technology, in a case resolved through settlement.

• Apple v. Rivos (2022): Apple sued chip startup Rivos over the hiring of several former Apple employees and the alleged misappropriation of trade secrets related to system-on-chip (SoC) design. The litigation was resolved through a settlement in 2024, the financial terms of which were not disclosed, as part of an agreement that allowed Apple to examine Rivos's systems to recover confidential information.

• xAI v. OpenAI (2025-2026): xAI, Elon Musk's company, sued OpenAI over alleged misappropriation of trade secrets related to its Grok AI model, but Judge Rita Lin dismissed the case in June 2026. This precedent, favorable to OpenAI though arising from different facts, could be invoked by the defense.

Procedural outlook for the present litigation

In light of the Waymo v. Uber precedent and Northern District of California case law, the procedural outlook for the present litigation can be summarized as follows:

1. Apple has a reasonable chance of obtaining a preliminary injunction if it can establish likelihood of success on the merits and irreparable harm. The allegations of file downloads, device retention, and use of supplier information provide a stronger factual foundation than mere suspicion.

2. OpenAI faces a significant procedural risk in discovery, where the true scope of the alleged misappropriation could come to light. OpenAI's failure to respond to Apple's February 2026 letter could be used against it.

3. A settlement is a plausible scenario, especially if the court grants an injunction disrupting OpenAI's hardware program. The Waymo v. Uber precedent suggests a monetary settlement could be the most rational outcome for both parties.

4. The litigation could drag on for years if either party appeals interlocutory rulings or the final judgment. However, the pressure of the IPO and hardware development could incentivize an early settlement.

Available sources do not permit a definitive conclusion as to which of these scenarios will materialize, nor do they specify whether the parties have entered settlement talks. The only certainty is that the litigation remains at an early stage, and that whatever its outcome, it will carry significant implications for the tech industry as a whole.

12. Conclusions

12.1. Summary of the parties' principal arguments

The complaint in Apple Inc. v. Liu et al., 5:26-cv-07078 (N.D. Cal. 2026), marks a milestone in the convergence of artificial intelligence and consumer hardware, placing at the center of the legal debate the protection of trade secrets within a collaboration-competition ecosystem. Analysis of the litigation's factual and legal foundations yields the following synthetic conclusions.

Apple's position is built around the allegation of a "coordinated pattern of institution-wide misconduct" by OpenAI, which allegedly used the recruiting process as a mechanism to extract confidential Apple information, including technical specifications, unreleased product designs, manufacturing processes, and supplier relationships. The company contends that the conduct of the individual employee defendants — Liu and Tan — was not isolated, but instigated and facilitated by OpenAI, which allegedly "normalized and exemplified" the misconduct at every level of the organization. Apple grounds its claim in the Defend Trade Secrets Act of 2016 (DTSA) and the California Uniform Trade Secrets Act (CUTSA), seeking injunctive relief, compensatory damages, exemplary damages, and costs. The company emphasizes that it adopted "reasonable measures" to preserve the confidentiality of its information, including confidentiality agreements, annual training, and security policies.

OpenAI's position, for its part, has so far been articulated as a categorical denial: "we have no interest in using anyone's trade secrets" and "we are not aware of any evidence supporting the merits of this lawsuit." The company maintains that it is "focused on building innovative technology" and defends employees' freedom to work wherever they choose. Although OpenAI has not yet filed a formal response to the complaint, its possible lines of defense include the absence of a trade secret, the absence of misappropriation, the absence of reasonable measures by Apple, and the lack of a causal link between the alleged misappropriation and the damages claimed. OpenAI's defense could also invoke California's rejection of non-compete agreements and the inevitable disclosure doctrine.

12.2. Assessing the strength of Apple's allegations in light of existing case law

Assessing the strength of Apple's allegations requires considering both the factual elements of the complaint and the legal and case-law framework within which the litigation unfolds.

Strengths of Apple's position:

1. Concrete, verifiable factual allegations: unlike other trade secret litigation grounded in mere suspicion, Apple's complaint incorporates specific factual allegations that, if proven, could constitute misappropriation: Liu's downloading of "dozens of confidential files" through a security flaw he exploited, retention of a company MacBook, coaching candidates to bring "real pieces" of Apple hardware to interviews, and use of supplier information to contact Apple's manufacturing partners. These allegations are stronger than mere circumstantial inference.

2. The existence of confidentiality agreements: Apple has entered into confidentiality agreements with its employees that survive termination of employment, a relevant element for establishing the adoption of "reasonable measures" and for supporting the breach-of-contract claims.

3. The Waymo v. Uber precedent: the litigation bears significant parallels to Waymo v. Uber, in which the plaintiff obtained an injunction and forced a favorable settlement through allegations of file theft and hiring away a competing team. This precedent suggests Apple has a reasonable chance of success in seeking injunctive relief.

4. Trade secret case law: federal and California case law has established that actively soliciting confidential information from a competitor's employees, as well as exploiting security vulnerabilities, constitutes "improper means" under both the DTSA and CUTSA.

Weaknesses and risks of Apple's position:

1. The distinctive character of California law: California rejects non-compete agreements and the inevitable disclosure doctrine, limiting Apple's ability to challenge OpenAI's mere hiring of its former employees. Apple cannot sue OpenAI for hiring more than 400 former employees, only for the specific conduct constituting misappropriation.

2. The burden of proof: Apple bears the burden of proving the existence of the trade secret, the adoption of reasonable measures, and the defendants' misappropriation. This burden is particularly demanding in trade secret matters, where the plaintiff must show the information is not generally known or readily ascertainable.

3. The risk that discovery will not confirm the allegations: discovery could reveal that the information Liu and Tan accessed was not truly secret, or that OpenAI did not misuse it. As Professor Mark Lemley has noted, "if Apple's claims that employees took confidential documents — and that OpenAI is using those documents — are true, that is indeed a problem for OpenAI." The conditional matters.

4. The xAI v. OpenAI precedent: Judge Rita Lin's recent dismissal of xAI's trade secret misappropriation lawsuit against OpenAI, though arising from different facts, shows that trade secret lawsuits do not always succeed. OpenAI could invoke this precedent to argue that Apple's allegations are equally speculative.

Overall assessment: Apple's complaint presents a factual foundation stronger than mere suspicion or the hiring of former employees, and is grounded in specific allegations that, if proven, would fit within the "improper means" concept under the DTSA and CUTSA. The litigation's success is not assured, however, and OpenAI has plausible lines of defense, particularly under California law. Discovery will be decisive in determining whether Apple's allegations are confirmed or whether, instead, OpenAI can show the information at issue was not secret or was not misused.

12.3. Broader implications for trade secret protection in the AI economy

The Apple v. OpenAI litigation transcends the dispute between two companies and raises implications of broad significance for the tech ecosystem and for trade secret protection in the AI economy.

1. The convergence of AI and hardware redefines the value of trade secrets

The complaint underscores that the next phase of the AI battle will be fought in hardware and physical devices. As analyst Paolo Pescatore has observed, "the AI race is moving beyond models and chatbots to who controls the device, the interface, and the direct relationship with the consumer." This convergence means that trade secrets related to industrial design, manufacturing processes, supplier relationships, and product roadmaps acquire value equivalent to that of algorithms. AI companies seeking to expand into hardware will need to build their own capabilities without resorting to established manufacturers' confidential information, or they will assume the risk of litigation like this case.

2. Labor mobility and trade secrets in tension

The case illustrates the structural tension between labor mobility — protected under California law — and trade secret protection. Professor Mark Lemley has noted that merely hiring a competitor's former employees "isn't illegal in California, where Silicon Valley itself grew up in part because of state laws that let employees leave for a competitor." However, stealing documents, retaining devices, and using confidential information can indeed be unlawful. The litigation could have a chilling effect on recruiting practices in the tech sector, leading companies to implement stricter controls and adopt more cautious hiring policies. The result could be a tech ecosystem that is "slower and more cautious."

3. The fragility of tech alliances

The evolution of the Apple-OpenAI relationship — from partners in 2024 to litigants in 2026 — illustrates the fragility of tech alliances when the parties' strategic interests converge toward direct competition. AI companies and hardware manufacturers that form alliances must be aware that collaboration can turn into competition, and that knowledge gained during the collaboration can become the subject of subsequent litigation. This risk could incentivize vertical integration or the negotiation of more restrictive confidentiality and non-compete agreements.

4. The impact on OpenAI's IPO and the AI bubble

The timing Apple chose to file suit — amid OpenAI's IPO preparations and the launch of its first hardware product — suggests the litigation also carries a strategic dimension. Legal uncertainty and the possibility of injunctive relief could cool investor appetite and delay or complicate OpenAI's path to going public. More broadly, the litigation could test the AI bubble, given that OpenAI is a central piece of the AI ecosystem and a participant in the U.S. government's Stargate project to secure American dominance in artificial intelligence. "There's so much at stake here that a misstep by OpenAI, which no longer enjoys a stellar reputation, could drag down a significant part of the AI ecosystem with it."

5. The need for an adapted legal framework

The litigation highlights the need for a legal framework that balances trade secret protection with the promotion of innovation and labor mobility. The DTSA and CUTSA provide tools for protecting trade secrets, but do not resolve the tension between employees' freedom to change jobs and their obligation to preserve the confidentiality of their former employers' information. Apple v. OpenAI could set a precedent for how courts should address this tension in the context of artificial intelligence and hardware.

Final conclusion

The complaint in Apple Inc. v. Liu et al., 5:26-cv-07078, is a highly consequential piece of litigation that places trade secret protection in the AI economy at the center of the legal debate. Apple's allegations present a solid factual foundation and rest on specific conduct that, if proven, would fit within the "improper means" concept under the DTSA and CUTSA. The litigation's success is not assured, however, and OpenAI has plausible lines of defense, particularly under California law. Discovery will be decisive in determining the outcome of the case, which could be resolved through settlement, injunctive relief, or trial. Whatever the outcome, the litigation will carry significant implications for the relationship between AI companies and hardware manufacturers, for labor mobility in the tech sector, and for the artificial intelligence ecosystem as a whole.

Sources and bibliography

Court documents

• Apple Inc. v. Chang Liu, Tang Yew Tan, OpenAI Foundation, OpenAI Group PBC and io Products, LLC, No. 5:26-cv-07078 (N.D. Cal., filed July 10, 2026). Docket available via CourtListener and PacerMonitor.
• Waymo LLC v. Uber Technologies, Inc., No. 3:17-cv-00939 (N.D. Cal. 2017), settled February 9, 2018.
• Whyte v. Schlage Lock Co., 101 Cal. App. 4th 1443 (Cal. Ct. App. 2002).
• Winter v. Natural Resources Defense Council, Inc., 555 U.S. 7 (2008).
• xAI Corp. v. OpenAI, Inc. (N.D. Cal.), order of dismissal with prejudice issued by Magistrate Judge Rita F. Lin on June 15, 2026.

Statutes

• Defend Trade Secrets Act of 2016, 18 U.S.C. § 1836 et seq.
• California Uniform Trade Secrets Act, Cal. Civ. Code § 3426 et seq.
• Federal Rules of Civil Procedure, Rules 12(b)(6) and 56.

Press and legal analysis

• Bloomberg, "Apple Sues OpenAI for Trade Secret Theft Over AI Hardware Designs," July 10, 2026.
• The Washington Post, "Apple sues OpenAI, alleging the AI company stole trade secrets," July 10, 2026.
• CNBC, "Apple sues OpenAI alleging trade secret theft, says scheme was 'at every level,'" July 10, 2026.
• Axios, "Apple sues OpenAI for trade secret theft," July 10, 2026.
• Fortune, "Apple accuses OpenAI, and former design star Jony Ive's io Products firm, of stealing hardware trade secrets in blockbuster lawsuit," July 10, 2026.
• Yahoo Finance / Bloomberg, "Apple is suing OpenAI over alleged trade secret theft," July 13, 2026.
• TechCrunch, "The wildest allegations in Apple's trade secrets lawsuit against OpenAI," July 13, 2026.
• HackerNoon, "Apple v. OpenAI Trade Secrets Lawsuit: The 42 Most Explosive Allegations & The Internet's Reactions," July 16, 2026.
• The Hill / Yahoo Finance, "Apple, OpenAI suit spotlights battle over physical AI" (comments from Paolo Pescatore, PP Foresight), July 14, 2026.
• Invezz, "Why Apple sued OpenAI: everything to know about the AI trade secrets lawsuit" (comments from Professor Mark Lemley, Stanford Law School), July 13, 2026.
• San Francisco Examiner, "OpenAI faces another bitter battle, now with Apple," July 15, 2026.
• Law360, docket sheet for Apple Inc. v. Liu et al., 5:26-cv-07078.
• CNBC, "OpenAI is buying iPhone designer Jony Ive's AI devices startup for $6.4 billion," and Built In San Francisco / Forbes / SiliconANGLE, coverage of the $6.5 billion io Products acquisition, May 2025.
• Al Jazeera / Courthouse News / Silicon.co.uk, coverage of Magistrate Judge Rita Lin's dismissal of xAI v. OpenAI, June 15-16, 2026.
• CNBC / Forbes, coverage of the Waymo v. Uber $245 million settlement, February 9, 2018.
• AppleInsider / Law360, coverage of the Apple v. Rivos settlement, 2024.

Methodological note: the facts, figures, dates, and direct quotations in this article have been cross-checked against publicly available news coverage and court filings as of July 20, 2026. Because the proceeding is at an early stage, the allegations described are unproven claims by the plaintiff, and procedural details (judicial assignment, discovery schedule) may change as the case progresses.