A federal judge has blocked a key move by X and xAI in their lawsuit against Apple and OpenAI over AI features on the iPhone. The companies wanted OpenAI to hand over its source code. The court refused. That decision removes one of the most aggressive discovery tools from the case and keeps OpenAI’s most sensitive material out of reach for now.
U.S. Magistrate Judge Hal R. Ray Jr. issued the order, which sits at the center of a growing pile of discovery disputes. He ruled that X and xAI did not meet the legal standard needed to force a company to disclose something as sensitive as its core code.
The judge said no
X and xAI argue that Apple’s integration of ChatGPT into iOS blocks rival AI products such as Grok from competing on equal terms. To support that claim, they asked for OpenAI’s source code, saying they needed it to test whether Grok could run inside Apple’s system.
OpenAI pushed back. It told the court that technical limits, not favoritism, explain why Grok does not integrate into Apple’s AI features. The judge sided with OpenAI on the discovery issue. He said the plaintiffs had not shown that they tried less intrusive ways to get the same information.
From the order:
“The Court concludes that OpenAI’s source code is not relevant to Plaintiffs’ claims and is not within the scope of discovery under Rule 26. Although OpenAI’s source code certainly would be of great interest to Plaintiffs, Rule 26 does not require its disclosure.”
The judge added that X and xAI still have other ways to test the feasibility of integrating Grok without gaining what he called “unfettered access” to OpenAI’s proprietary systems.
The court rejected a pressure tactic
X and xAI also tried another angle. They argued that if OpenAI refused to turn over the code, that refusal should count as proof that Grok could, in fact, work inside Apple’s platform.
The judge rejected that idea.
From the ruling:
“Plaintiffs present their competitor OpenAI with a choice: hand over its most sensitive proprietary information or admit that Grok could have been integrated into the iPhone operating system. The Court does not order OpenAI to produce its source code.”
In short, the court refused to let one side force the other into a technical concession by threatening exposure of trade secrets.
A pattern of discovery fights
The judge also pointed to how fast the case has turned into a procedural brawl. Even though the lawsuit is less than five months old, the docket already includes more than 135 entries, many of them tied to discovery disputes. That history shaped the tone of the ruling and showed that the court has little patience for broad fishing expeditions.
Another recent setback fits the same pattern. South Korea’s government denied X and xAI’s request for documents from Kakao, a major “super app” operator, after finding the scope too broad and not proportional to the needs of the case.
The lawsuit continues. This order only blocks one path that X and xAI wanted to use. Still, it sets a clear boundary. Courts protect source code more than almost any other business record. If the plaintiffs want to keep pushing their claims, they will need to rely on documents, expert testimony, and technical evidence that do not require cracking open OpenAI’s core systems.
