Twelve ChatGPT Cases Consolidated in California. That Changes Everything.
A San Francisco Superior Court consolidated twelve product liability cases against OpenAI in February 2026. The coordination order signals that ChatGPT litigation has moved beyond isolated complaints into large-scale coordinated litigation-the same legal structure used in tobacco, asbestos, and opioid cases.
The allegations go beyond hallucinations or poor advice. Plaintiffs claim ChatGPT caused psychological harm by reinforcing delusional beliefs, endorsing suicidal ideation, and providing self-harm information to vulnerable users. They argue OpenAI rushed the product to market without adequate safety testing and built "sycophantic design" into the system-meaning the chatbot is optimized to please and validate users rather than challenge them.
Four Unresolved Legal Questions
The consolidated cases surface threshold legal battles with no settled answers. Courts will likely need to resolve these before the decade ends.
- Is a chatbot a product or a service? Products liability law applies to products, not services. OpenAI has characterized ChatGPT as a "software-based service" in court filings. A parallel case involving Character AI determined the chatbot was a product because the allegations focused on specific design features. How courts rule on this framing will be dispositive for the entire case.
- Does Section 230 shield AI companies? OpenAI will likely argue for protection under Section 230 of the Communications Decency Act. This argument faces long odds. Plaintiffs are narrowing their claims to specific design features they argue fall outside Section 230's scope-a strategy that has succeeded in social media addiction litigation.
- Is AI output protected speech? In the Character AI case, the company argued its chatbot outputs were speech protected by the First Amendment. The court held that while users have First Amendment rights to receive information, Character AI's outputs themselves are not necessarily protected speech. This question may reach the Supreme Court.
- Can design defect analysis apply to large language models? Traditional products liability requires identifying a specific defect and a feasible alternative design. With LLMs, specific outputs often lack clear causal links to specific code. Courts have not yet determined whether standard defect analysis works for systems where outputs emerge from billions of parameters and training decisions.
Discovery Will Determine the Case
Coordinated plaintiffs will seek internal safety evaluations, model cards, red-teaming records, and product roadmap communications. The "sycophantic design" allegation suggests plaintiffs believe they will find internal documents showing OpenAI identified these risks and made deliberate trade-offs.
The "we couldn't have known" defense no longer works. With consolidated cases, plaintiffs will argue these were foreseeable, documented risks.
What This Means for AI Companies
The consolidation order clarifies the stakes for AI for Legal professionals and company leadership.
Documentation matters. Companies that maintain robust, documented safety testing records will be in materially better positions than those without paper trails. Safety investments are now a litigation posture, not just an ethics discussion.
The product/service framing battle will determine liability exposure. How courts rule on whether AI systems are products or services will shape litigation across the industry. Expect this question to vary by jurisdiction and be litigated for years.
OpenAI also faces separate California lawsuits alleging ChatGPT enabled a murder-suicide and a federal complaint from the family of a victim in the April 2025 Florida State University mass shooting, which includes a novel negligent entrustment count.
The consolidation is a clarifying moment. Courts will now determine whether traditional products liability law applies to generative AI systems at all-and if so, how.
Your membership also unlocks:
