IP Law Daily, TRADE SECRETS—N.D. Cal.: OpenAI granted dismissal of trade secret, unfair competition claims by competitor, (Feb 25, 2026)

By Patricia K. Ruiz, J.D.

While the First Amended Complaint may state claims against former employees based on their alleged conduct, it does not state a plausible misappropriation claim against OpenAI as the sole defendant.

The U.S. District Court for the Northern District of California granted a motion to dismiss by OpenAI, Inc., with leave to amend trade secret and unfair competition claims by xAI Corp. The court held that the First Amended Complaint (FAC) does not plead nonconclusory facts showing OpenAI acquired, used, disclosed, or induced the misappropriation of xAI’s trade secrets. The court emphasized that while several former xAI employees allegedly exfiltrated information before leaving for OpenAI, the FAC does not point to any misconduct by OpenAI and fails to plausibly allege direct, indirect, or vicarious misappropriation under the Defend Trade Secrets Act (DTSA) (X.AI Corp. v. OpenAI, Inc., No. 3:25-cv-08133-RFL (N.D. Cal. Feb. 24, 2026)).

Background. xAI and OpenAI are competitors in generative AI. Eight xAI employees departed for OpenAI during the summer of 2025, including engineer Xuechen Li and London based engineer Jimmy Fraiture. As alleged, Li uploaded xAI’s source code to a personal cloud account connected to ChatGPT while communicating with an OpenAI recruiter, and Fraiture copied source code and a confidential all hands video to a personal device shortly after accepting an OpenAI offer. However, the court found the FAC lacks facts showing OpenAI directed, encouraged, or used any of that information. The order notes OpenAI later revoked Li’s offer following a temporary restraining order that limited his role, and that the FAC does not allege subsequent use of xAI information by Fraiture once employed at OpenAI.

Misappropriation. Applying Rule 12(b)(6) and the DTSA’s definitions of misappropriation, the court held xAI pleaded neither direct misappropriation (including inducement) nor indirect misappropriation (knowledge-based acquisition, disclosure, or use). On inducement, the court concluded that overlapping recruiter contacts, encrypted messaging, and timing around alleged exfiltration “without more” do not support a reasonable inference that OpenAI encouraged theft, distinguishing cases where defendants allegedly conditioned employment benefits on or otherwise orchestrated the taking of specific confidential assets. On indirect misappropriation, the court found no facts to infer OpenAI knew Li’s presentation contained xAI trade secrets or that any information was improperly acquired or mistakenly disclosed.

Vicarious liability. The court also rejected vicarious liability under respondeat superior, explaining that liability requires alleged use of trade secrets within the scope of employment after the hire—something the FAC does not supply. The court reiterated that “mere possession” does not constitute misappropriation.

UCL claims. Turning to xAI’s California Unfair Competition Law (UCL) claims, the court dismissed both the “unlawful” and “unfair” prongs as preempted by the California Uniform Trade Secrets Act (CUTSA). The “unlawful” prong rose and fell with the DTSA theory, and the “unfair” prong—framed as anticompetitive harm from OpenAI’s purported scheme to neutralize xAI’s innovations—was, as pleaded, rooted in alleged trade secret misappropriation and thus displaced by CUTSA. The court noted xAI’s “poaching” allegations were presented as a means to access xAI’s confidential information and did not independently allege anticompetitive harm untethered to trade secret claims.

The court recognized that circumstantial evidence can support inferences in trade secret cases but stressed that Rule 12(b)(6) standards do not relax in this context. The FAC’s “gestalt” of allegations, or, in other words, allegations as a whole, still failed to plausibly allege OpenAI’s use, acquisition with the requisite knowledge, or inducement. The court cited contrasting matters where defendants rapidly released products mirroring protected functionality or where there was concrete linkage between the defendant’s outputs and the plaintiff’s alleged secrets—linkage absent here.

The Case is No. 3:25-cv-08133-RFL.

Judge: Lin, R.

Attorneys: Leelle Bruerea Slifer (Winston & Strawn LLP) for X.AI Corp. Carolyn Hoecker Luedtke (Munger, Tolles Olson LLP) for OpenAI, Inc.

Companies: X.AI Corp.; OpenAI, Inc.

Cases: TradeSecrets AINews CaliforniaNews GCNNews