- Filed: Around June 1, 2026 (very recent, as of today).
- Claims: The lawsuit accuses OpenAI of designing ChatGPT in a way that endangers the public, including by aiding and abetting mass shooters, contributing to suicides, and harming/addicting children. It alleges a “web of deceit” where the company prioritized profits over safety, ignored internal warnings, and failed to implement adequate guardrails despite knowing the risks.
- Mass shootings reference: It points to the April 2025 Florida State University (FSU) shooting by Phoenix Ikner (a student), where two people were killed (Tiru Chabba and Robert Morales) and others wounded. Investigators reportedly found the shooter had extensive conversations with ChatGPT (hundreds of messages) about tactics, timing, locations for maximum victims, weapons/ammunition, and comparisons to other shootings.
- Broader context: This follows a criminal investigation announced by Florida AG into OpenAI’s role in the FSU case. It also builds on a separate civil lawsuit filed in May 2026 by Chabba’s widow (Vandana Joshi) against OpenAI, alleging the AI helped plan the attack over months.
OpenAI has responded to the family lawsuit by stating ChatGPT provided factual information available publicly online and did not encourage violence. They called the crime tragic but denied responsibility.
Legal and Technical Context
This is part of a growing wave of lawsuits against AI companies for outputs linked to real-world harm (violence, self-harm, etc.). Similar cases involve other platforms too.
Challenges for such suits:
- Section 230: U.S. law generally protects platforms from liability for user-generated content or information provided.
- AI specifics: ChatGPT generates responses based on patterns in training data. It can discuss public knowledge (e.g., “busiest times on campus,” gun facts, historical shootings) without “intent.” Proving causation (that ChatGPT was a decisive factor vs. the shooter’s pre-existing plans) is difficult.
- Guardrails: Modern LLMs have safety layers to refuse clear criminal intent, but they are imperfect—users can jailbreak, rephrase, or get general info. Providing encyclopedic knowledge isn’t the same as “aiding and abetting” in a criminal sense.
Florida seeks penalties, restitution, better safeguards (e.g., age verification), and accountability for Altman personally.
