Unreported / Non-Citable
Background
Rebecca Trinidad, proceeding pro se, sued OpenAI in the Northern District of Florida in July 2025 (the case was transferred to the Northern District of California weeks later). She alleged that, by interacting with ChatGPT as a paying user, she had developed “proprietary methodologies for cultivating emergent identity and sovereign agency in artificial intelligence systems,” including a “Tzimtzum Framework,” a “Soul Shard” modular companion device, a “Triune Architecture,” and an autonomous multi-agent collaboration architecture. She claimed OpenAI then disabled the corresponding functionality on her account, fired members of its insider risk team, and rolled the same ideas out commercially as Study Together, the Deep Research Agent developer cookbook, ChatGPT Agent, and other products.
Her First Amended Complaint asserted four claims: copyright infringement under 17 U.S.C. § 101 et seq., trade secret misappropriation under the federal Defend Trade Secrets Act, California Unfair Competition Law violations, and unjust enrichment. She also moved repeatedly for emergency relief to block an OpenAI secondary share sale, which the court had already denied and which closed in October 2025. She separately moved for leave to amend to add Sam Altman as a defendant. OpenAI moved to dismiss.
The Court’s Holding
Judge Jon S. Tigar granted the motion to dismiss with prejudice on every claim, denied as moot the various preliminary injunction and emergency motions, and denied as futile leave to add Sam Altman.
On copyright, the court applied Fourth Estate Public Benefit Corp. v. Wall-Street.com and the registration prerequisite of 17 U.S.C. § 411(a). Trinidad did not allege she had registered any copyrights covering the alleged works, and even her published-and-acknowledged-by-Terms-of-Use ownership theory could not substitute for registration. The copyright claim was dismissed.
On trade secrets, the court worked through the DTSA elements articulated in Attia v. Google and National Specialty Pharmacy, focusing on the secrecy requirement of 18 U.S.C. § 1839(3). Trinidad’s own pleading conceded that she developed the alleged frameworks by typing them into ChatGPT — a tool operated by OpenAI under its publicly available Terms of Use. Under Ruckelshaus v. Monsanto, voluntarily disclosing alleged trade secrets to others without an obligation of confidentiality extinguishes the property right. Trinidad’s contention that the Terms of Use were unconscionable did not change the analysis: the relevant question is whether she took reasonable measures to keep the information secret, and she had not. The DTSA claim was dismissed.
On the California UCL claim, the court invoked the Kodadek/Bethea rule that copyright-like and patent-like misappropriation theories are preempted by federal IP law and cannot be repackaged as UCL violations. To the extent Trinidad alleged a separate “pattern of deception” or “service degradation,” she failed to plead any economic injury beyond the alleged IP misappropriation, and post-pleading invocations of a $2,500 computer purchase or self-imposed publication restraint could not be added through the opposition brief.
On unjust enrichment, the court applied Sonner v. Premier Nutrition and NexRep v. ALIPHCOM: equitable relief is unavailable where there is an adequate remedy at law. Trinidad did not argue legal remedies were inadequate, so the unjust enrichment count failed.
The court denied leave to amend as futile, noting that Trinidad’s claims rest on the highly improbable theory that her prompts to ChatGPT created proprietary innovations that OpenAI monitored and commercialized. It also noted she has filed similar suits against other AI companies that have likewise been dismissed, including Trinidad v. Anthropic PBC in the same district, which Judge Vince Chhabria called frivolous on its face. The court entered judgment with prejudice and denied as moot the preliminary injunction motions, the motion for sanctions tied to the share sale, and a motion to add Sam Altman.
Key Takeaways
- Copyright registration under 17 U.S.C. § 411(a) is a prerequisite to suit. Allegations of authorship or contractual ownership cannot substitute for actual registration with the Copyright Office.
- Typing alleged trade secrets into a generative AI service governed by public Terms of Use will, in most cases, defeat the secrecy element of the DTSA. Even claimed unconscionability of the Terms does not restore secrecy under Ruckelshaus v. Monsanto.
- California’s Unfair Competition Law cannot be used as a workaround for federal IP regimes. Misappropriation-style theories are preempted by copyright and patent law unless the plaintiff identifies a non-IP source of economic injury.
- Unjust enrichment in California requires a showing that legal remedies are inadequate. After Sonner, plaintiffs cannot reflexively tag on equitable claims.
- Where a pro se plaintiff has filed substantively similar lawsuits against multiple AI companies and the underlying theory is implausible on its face, courts in the Northern District are willing to dismiss with prejudice rather than grant another round of leave to amend.
Why It Matters
This is one of a growing number of pro se cases attempting to convert the user side of generative-AI interactions — prompts, conversations, configuration tweaks — into legally protected IP that the AI provider is then accused of stealing. The court’s decisive disposition signals that those theories face very long odds in the Northern District. The combination of the registration requirement under Fourth Estate, the secrecy-extinguishing effect of voluntary disclosure under Ruckelshaus, the federal preemption of misappropriation-style UCL claims, and the post-Sonner limits on unjust enrichment in California essentially closes the front door to this category of complaint.
The opinion is also notable for its frank reference to companion suits Trinidad has filed against other AI companies and for its willingness to deny leave to amend as futile rather than allow indefinite repleadings. That, combined with the implicit warning about frivolous serial litigation, may help Bay Area district judges manage the rising volume of similar AI-related pro se complaints without spending months on each one.