OpenAI asked a federal judge on August 6, 2026, to dismiss Apple's trade secrets lawsuit, filing a motion in the U.S. District Court for the Northern District of California that calls the allegations "meritless" (The Verge; Reuters). The case, docketed as Apple Inc. v. Liu, No. 5:26-cv-07078, was filed July 10, 2026. It accuses two former Apple employees now working at OpenAI of misappropriating trade secrets — that is, taking confidential proprietary information without permission — to advance OpenAI's hardware development plans (CourtListener).

In its dismissal motion, OpenAI advances three main arguments. First, that Apple mischaracterized the employees' actions as theft. Second, that the information Apple calls "trade secrets" is really "generic" product development information rather than material that legally qualifies for trade secret protection. Third, that Apple failed to take reasonable steps to keep the information secret, which is a required element under trade secret law (The Verge). The motion also alleges that Apple's case is poorly investigated, built on selective or out-of-context communications, and is "rotten to its core" (The Verge).

One of the two accused employees, Chang Liu, is alleged to have downloaded confidential files from Apple after his departure. OpenAI counters that Liu was simply helping former colleagues who had asked him for assistance, framing the activity as benign rather than misappropriation (The Verge). The dismissal motion was filed under case number gov.uscourts.cand.474095, and the court filing is publicly available via CourtListener's docket repository (CourtListener).

OpenAI has not confined its response to the courtroom. On Tuesday, August 4, the company published a blog post titled "Apple is getting this wrong," describing the lawsuit as "careless, aggressive, and oddly personal" (The Verge). The post signals that OpenAI intends to contest this matter publicly as well as legally, a posture that distinguishes it from the typical quiet defensiveness companies adopt when facing litigation from a peer.

Two days before OpenAI's dismissal motion, on August 4, Apple asked the court for a preliminary injunction — a court order that would bar OpenAI and the two named former employees from accessing, acquiring, using, or disclosing the alleged confidential information while the lawsuit proceeds (Reuters). The injunction request raises the stakes: if granted, it would impose restrictions on OpenAI's hardware operations during what could be a protracted discovery and litigation timeline, potentially constraining the company's access to personnel or knowledge relevant to those efforts. The injunction filing is also part of the docket at gov.uscourts.cand.474095 (CourtListener).

A judge is set to hear arguments on OpenAI's motion to dismiss on October 1 (The Verge).

The dispute centers on a familiar pattern in the technology sector: employees leave one company for a competitor, and the former employer claims proprietary knowledge traveled with them. Trade secret litigation of this kind has historically turned on three questions: whether the information at issue qualifies as genuinely proprietary, whether the employer took adequate steps to protect it, and whether the alleged misappropriation resulted in actual or threatened competitive harm. OpenAI's motion targets each of these elements directly. The argument that Apple's information is "generic" and that secrecy measures were inadequate goes to the threshold question of whether protectable trade secrets exist at all. If the court accepts that framing at the motion-to-dismiss stage, the case could end before discovery — the pre-trial phase where both sides exchange evidence — meaningfully begins.

Apple's injunction request operates on a different timeline and standard. Courts weigh preliminary injunctions using a sliding-scale test that considers the likelihood of success on the merits, the balance of harms to each side, and the public interest. Apple's burden here is to show not only that it is likely to prevail but that irreparable harm will result without the injunction. OpenAI's contention that the underlying conduct was benign, if credited by the court, would undercut both the likelihood-of-success and irreparable-harm prongs simultaneously.

The rhetorical intensity on both sides is notable. OpenAI's blog post and its motion's language, "rotten to its core," escalate well beyond the measured tone typical of early-stage litigation filings. Apple's decision to name individual employees, rather than suing OpenAI alone, signals a willingness to target the personnel dimension directly. Both choices suggest that neither party views this as a routine matter to be quietly settled.

The broader context here is worth separating from the day-to-day filings. For OpenAI, the lawsuit arrives as the company pursues hardware development, a domain where Apple's accumulated expertise in silicon design, system integration, and supply chain management sets a high bar. For Apple, the case tests how aggressively it can police the movement of talent and knowledge into a company that, until recently, operated primarily as a software and services provider. The October 1 hearing will offer the first signal of whether the court finds OpenAI's threshold challenges to the case persuasive or whether the dispute will proceed into full discovery.