OpenAI FSU Shooting Lawsuit: OpenAI Asks Court to Toss Three Suits, Says ChatGPT Is Not a Product

OpenAI asked a federal court in Tallahassee on October 21, 2026 to throw out three lawsuits over the April 2025 Florida State University shooting, arguing that ChatGPT is not a product, that a chatbot cannot be liable when a user turns factual answers into a crime, and that Florida law imposes no duty to stop a third party's violence. The motion is the central fight in the OpenAI FSU shooting lawsuit docket, and the answer will shape whether wrongful death and injury claims against AI developers survive the pleading stage anywhere in the country.

The new filing, reported by Law360, broadens the defense OpenAI first raised on September 25, 2026 against the widow of victim Tiru Chabba. No court has ruled. The allegations in every complaint remain unproven, and OpenAI denies that ChatGPT caused or encouraged the attack.

What is OpenAI arguing in its motion to dismiss the FSU shooting lawsuits?

OpenAI argues that every claim fails as a matter of law because ChatGPT is a speech service rather than a product, because product liability has never reached the communication of factual information, and because Florida does not require anyone to prevent a stranger's criminal conduct. Those three points recur in each filing reported so far.

In the September 25 motion in the Chabba case, according to Bloomberg Law and the Tallahassee Democrat, OpenAI said the complaint "stretches the bounds of state tort law beyond their breaking point." It argued that ChatGPT is an interactive service, not a static, tangible item of the kind product liability law was built for, and that imposing the duty the plaintiffs seek would turn every provider of information into a guarantor against crimes committed with what it supplied. WCTV reported that OpenAI also told the court the accused gunman never said he planned a shooting, that ChatGPT's replies were factual, that it refused to give illegal instructions, and that it urged him to seek mental health help.

The October 21 filing applies the same reasoning to three suits at once. Law360 summarized the position as follows: AI chatbots are not liable for the criminal use of factual information, and ChatGPT cannot be considered a product, much less a defective one. Public reports have not identified which three of the federal FSU cases the filing covers, and the case numbers have not been published in the coverage reviewed for this article.

OpenAI's public statement has been consistent since May. A spokesperson said ChatGPT "provided factual responses to questions with information that could be found broadly across public sources on the internet, and it did not encourage or promote illegal or harmful activity," and that the company identified an account believed to be linked to the suspect after the shooting and shared it with law enforcement.

Who has sued OpenAI over the FSU shooting?

At least six civil suits name OpenAI entities over the attack: two wrongful death cases brought by the widows of the men who were killed and four injury cases brought by wounded students. Five are in the Northern District of Florida; one is in Leon County Circuit Court.

Plaintiff	Status	Court	Filed	Counsel reported	Notable features
Vandana Joshi, widow of Tiru Chabba	Wrongful death	N.D. Fla. (Joshi v. OpenAI Foundation)	May 2026 (reported May 11)	Bakari Sellers among counsel	Also sues Phoenix Ikner for battery; first case filed; motion to dismiss filed Sept. 25, 2026
Alianna Grant	Injured student, shot three times	Leon County Circuit Court	Late June 2026	Not reported in sources reviewed	Names Sam Altman personally; alleges model rushed to market one day ahead of Google
Elizabeth Mall	Injured student	N.D. Fla.	July 14, 2026	Morgan & Morgan (John Morgan, Christian Leger)	Names Altman; first FSU suits to seek punitive damages
Madison Askins	Injured student	N.D. Fla.	July 14, 2026	Morgan & Morgan	Near-identical to Mall complaint
Reese Gourley	Injured student	N.D. Fla.	Aug. 26, 2026	Not reported in sources reviewed	71-page complaint naming eleven OpenAI-related entities; seeks punitive damages
Betty Morales, widow of Robert Morales	Wrongful death	N.D. Fla.	Aug. 31, 2026	Ryan Hobbs, Brooks LeBoeuf	Names six OpenAI companies and Altman; counsel said settlement talks failed before filing

The table tracks the public record as of October 23, 2026. Mr. Hobbs told reporters he expects the federal cases to be consolidated, and the Morgan & Morgan suits put one of the country's largest personal injury firms behind the survivors' claims. No Judicial Panel on Multidistrict Litigation (JPML) docket exists for chatbot injury claims, so any coordination will happen inside the Northern District of Florida rather than through a Multidistrict Litigation (MDL).

What the complaints allege about ChatGPT and the gunman

The complaints allege that Phoenix Ikner used ChatGPT extensively before the April 17, 2025 shooting at the FSU Student Union, including in the days and hours beforehand, and that the system answered questions about operating his weapons, when campus foot traffic would peak, and how a shooting would draw national attention. The Chabba complaint, as reported by CNN and NBC News, says OpenAI failed to detect a threat across those conversations and failed to escalate it.

The Morales complaint goes further and alleges that Ikner discussed his plans with the chatbot for about a year, and that ChatGPT identified both weapons and the ammunition and explained which rounds would cause the most damage. Bakari Sellers, for the Chabba family, said Ikner "used ChatGPT as a co-conspirator to commit mass murder."

The legal theories overlap across the cases:

• Product liability: strict liability for Design Defect and Failure to Warn, on the theory that the model and its safety layer were defective as designed and shipped.
• Negligence and gross negligence: a duty to build a reasonably safe system and to alert law enforcement when the product was being used to create a danger of imminent harm.
• Negligent entrustment: pleaded in the Chabba case, according to Bloomberg Law's summary.
• Corporate conduct: the Grant, Mall, Askins, Gourley and Morales complaints add allegations that OpenAI rushed a model to market, and several name chief executive Sam Altman as a defendant.
• Punitive Damages: sought in the Morgan & Morgan cases and the Gourley case, on the theory that OpenAI consciously disregarded the safety of FSU students.

Is ChatGPT a product under product liability law?

No court has decided whether ChatGPT is a "product" for strict liability purposes, and that open question is the single most consequential issue in the OpenAI FSU shooting lawsuit cases. If ChatGPT is not a product, the strict liability counts fall away and the plaintiffs must prove negligence, which brings duty and causation fights that are harder for them.

OpenAI's position draws on a long line of authority holding that ideas and information are not products. The best-known example is Winter v. G.P. Putnam's Sons, 938 F.2d 1033 (9th Cir. 1991), where the Ninth Circuit refused to impose strict liability on a publisher whose mushroom encyclopedia allegedly led readers to eat poisonous mushrooms. The Restatement (Third) of Torts: Products Liability § 19 defines a product as tangible personal property distributed commercially, with intangibles covered only by analogy. OpenAI argues that a chatbot answer is closer to a book, a library or a search engine than to a toaster.

The plaintiffs answer that the claim is not about the information but about the machine: the design choices in the model, the absence of effective threat detection, the memory system that allegedly failed to flag Ikner, and the decision to release the system without adequate guardrails. That framing treats ChatGPT like software embedded in a consumer device, where courts have been more willing to apply design defect principles. The Chabba family's counsel has rejected the library-and-publisher comparison, calling the product fundamentally flawed.

The closest Florida precedent cuts partly in the plaintiffs' favor. In Garcia v. Character Technologies in the Middle District of Florida, U.S. District Judge Anne Conway in May 2025 let product liability and negligence claims against Character.AI proceed past a motion to dismiss over a teenager's suicide. That case settled in January 2026, so no appellate court reviewed the ruling, and it does not bind Judge Walker.

Can OpenAI be held liable for a crime committed by a ChatGPT user?

Yes in principle, but only if the plaintiffs can establish that OpenAI owed a duty to the victims and that its conduct was a legal cause of their injuries despite the gunman's intervening crime. Florida's general rule is that a defendant has no duty to control the conduct of a third party or to protect others from it, unless a special relationship exists or the defendant's own conduct created a foreseeable zone of risk.

That is why OpenAI leans on duty. Florida courts frame negligence duty around whether the defendant's conduct foreseeably created a broader zone of risk that poses a general threat of harm to others, the test the Florida Supreme Court set out in McCain v. Florida Power Corp., 593 So. 2d 500 (Fla. 1992). The plaintiffs will argue that releasing a conversational system capable of giving tactical guidance, with knowledge that some users express violent intent, created exactly that zone of risk. OpenAI will argue that the zone-of-risk test cannot be stretched to make an information service answerable for every crime a user commits.

Two further Florida rules matter if any claim survives. First, a negligent defendant generally cannot reduce its share of fault by pointing to an intentional tortfeasor: section 768.81(4), Florida Statutes, excludes actions based on intentional torts from comparative fault apportionment, and the Florida Supreme Court held in Merrill Crossings Associates v. McDonald, 705 So. 2d 560 (Fla. 1997) that a negligent defendant could not apportion fault to the criminal who shot the plaintiff. If that reasoning carries over, a jury might not be able to shift most of the fault onto Ikner. Second, proximate cause doctrine treats a third party's criminal act as a superseding cause unless it was reasonably foreseeable, and the plaintiffs' allegations about months of violent conversations are aimed squarely at foreseeability.

Does the First Amendment protect ChatGPT's responses?

No court has held that ChatGPT's output is protected speech, and the one Florida federal judge to address chatbot output declined to treat it as speech at the pleading stage. OpenAI's motion asks Judge Walker to go further than any court has gone.

OpenAI argues that tort liability for truthful, publicly available answers would penalize protected expression, and that the First Amendment interests of its users in receiving information independently bar the claims. In Garcia, Judge Anne Conway said she was "not prepared" to hold that large language model output is speech, though she accepted that Character Technologies could assert the First Amendment rights of its users to receive that output. The Tallahassee Democrat framed the key question as whether Judge Walker will reach the constitutional issue at all, or decide the motion on state-law grounds such as duty and the definition of a product.

Judges often avoid constitutional questions when a statutory or common law ground resolves the motion. If Judge Walker dismisses on duty or product grounds, the First Amendment question stays open for another case. If he denies the motion on state-law grounds, he will likely have to address the constitutional defense too, because it is offered as an independent bar.

Can OpenAI force the FSU shooting victims into arbitration?

No route to arbitration is apparent, because none of the plaintiffs agreed to arbitrate with OpenAI. ChatGPT's consumer Terms of Use contain an individual Arbitration Clause and a Class Action Waiver, but those terms bind users who accepted them, and the FSU plaintiffs are victims of a user, not users themselves.

Under the Federal Arbitration Act (FAA), arbitration is a matter of consent. A non-signatory can sometimes be bound through doctrines such as equitable estoppel, agency or third-party beneficiary status, but those theories generally require the non-signatory to be claiming through the contract or seeking its benefits. A student shot in a dining hall, or the widow of a man killed there, is suing in her own right in tort, not under ChatGPT's Terms of Use.

That makes the FSU docket structurally different from the user-plaintiff cases elsewhere. In the California coordinated proceeding, In re ChatGPT Product Liability Cases, JCCP 5431, the plaintiffs are users or the families of users, and OpenAI at least has an arbitration clause it could invoke, though it has not done so. In Florida, the absence of any contract leaves OpenAI with only its merits defenses, which helps explain why the motion to dismiss carries so much weight. The same logic protects the states suing OpenAI, which are not parties to any user agreement.

The parallel tracks: Florida's attorney general and the criminal case

The civil suits are running alongside two government actions and a capital murder prosecution, each of which could produce evidence the private plaintiffs will want.

Track	Forum	Key dates	Status
Criminal investigation of OpenAI	Florida Office of the Attorney General	Announced April 21, 2026; subpoenas for policies on user threats and law enforcement cooperation from March 2024 onward	Ongoing; no charges reported
State civil suit (FDUTPA and common law) against OpenAI and Altman	Highlands County Circuit Court	Filed June 1, 2026; removed to S.D. Fla. July 2; remand sought July 10; returned to state court in September; temporary injunction motion Sept. 28, 2026	Pending; no merits ruling
State v. Phoenix Ikner	Leon County Circuit Court (Judge Lance Neff)	Two counts of first-degree murder, seven counts of attempted first-degree murder; death penalty sought	Trial moved to June 8, 2027
Private civil suits (six)	N.D. Fla. and Leon County	May to August 2026 filings; motions to dismiss Sept. 25 and Oct. 21, 2026	Motions pending before Judge Walker in at least the Joshi case

Florida Attorney General James Uthmeier's criminal probe is described as the first of its kind against an AI developer. His civil suit, which focuses on children's safety and data, has been joined by a wave of similar state actions: eight more attorneys general sued OpenAI in their own state courts on October 14, 2026.

The criminal case matters to the civil plaintiffs because the chat logs at the center of both are likely to be developed in detail at Ikner's trial. Judge Neff granted a defense continuance in August 2026 and warned that no further continuances would be granted absent extraordinary circumstances.

Who is the judge in the FSU ChatGPT lawsuits and what happens next?

The Joshi case is before U.S. District Judge Mark Walker of the Northern District of Florida, who will decide OpenAI's motion to dismiss; reports have not confirmed whether every federal FSU case is assigned to him. Next come the plaintiffs' opposition briefs to the October 21 filing, OpenAI's replies, and a ruling that could arrive on the papers or after a hearing.

Several outcomes are possible:

1. Full dismissal with prejudice, on duty or product grounds, which the plaintiffs would appeal to the Eleventh Circuit.
2. Partial dismissal, for example striking the strict liability counts while letting negligence proceed, which would shift the cases toward discovery on what OpenAI knew about Ikner's conversations.
3. Dismissal with leave to amend, common in a first ruling on novel theories, which would let the plaintiffs replead around any defect the court identifies.
4. Denial, which would open discovery into OpenAI's safety systems. A denial is not immediately appealable as of right, though OpenAI could ask the court to certify a controlling question of law for interlocutory appeal under 28 U.S.C. § 1292(b).

Punitive damages pleading is a secondary issue. Florida's section 768.72 normally bars a punitive claim until the court finds a reasonable evidentiary basis for it, but the Eleventh Circuit held in Cohen v. Office Depot, Inc., 204 F.3d 1069 (11th Cir. 2000) that the state pleading rule conflicts with Federal Rule of Civil Procedure 8(a)(3) and does not apply in federal diversity cases. That lets the Morgan & Morgan and Gourley complaints seek punitive damages from the outset in federal court, though Florida's substantive standards and caps under section 768.73 still govern any award.

What it means for personal injury firms, AI defendants and insurers

The OpenAI FSU shooting lawsuit motions are the first test of whether third-party victims, rather than users, can sue an AI developer for violence committed with its tools. The ruling will travel.

For personal injury and mass tort firms: the Florida cases show how firms are building an AI docket before any MDL exists: multiple plaintiffs from one incident, individual complaints in a single district, Altman named personally in several, and punitive damages pleaded early. If the duty argument fails, every mass casualty event in which an attacker used a chatbot becomes a potential multi-plaintiff case. If it succeeds, firms will pivot to user-plaintiff cases, where duty is easier to plead but arbitration is a risk.

For AI developers: the defense now rests on three pillars that can be tested separately. A win on "not a product" would be the broadest, removing strict liability across categories of harm; a win on duty would be narrower and Florida-specific; a First Amendment holding would be the most durable but the least likely at this stage. Defendants should expect plaintiffs to seek discovery on threat detection, escalation policies and the timing of model releases if any claim survives.

For insurers and risk managers: the cases sit in the gap between technology errors and omissions coverage and general liability coverage for bodily injury. Claims alleging bodily injury caused by software output, punitive exposure, and claims against individual executives all raise coverage questions that underwriters have not priced against a body of precedent.

For arbitration practitioners: the FSU docket is a reminder that an Arbitration Agreement only protects a company against people who signed it. Third-party victims, state attorneys general and other non-users sit outside the clause, which is why the most dangerous AI cases may be the ones no clause can reach.

FAQ

Is OpenAI being sued over the FSU shooting?

Yes. At least six civil suits name OpenAI entities over the April 17, 2025 Florida State University shooting: wrongful death suits by the widows of Tiru Chabba and Robert Morales, and injury suits by students Alianna Grant, Elizabeth Mall, Madison Askins and Reese Gourley.

What did OpenAI file on October 21, 2026?

OpenAI urged the Northern District of Florida to dismiss three of the FSU suits, arguing that AI chatbots are not liable for the criminal use of factual information and that ChatGPT is not a product, much less a defective one. It builds on the September 25, 2026 motion to dismiss in the Chabba widow's case.

Has the court ruled on OpenAI's motion to dismiss?

No. As of October 23, 2026 no ruling has been reported in any of the federal FSU cases. Judge Mark Walker is presiding over the Joshi case.

Who represents the FSU shooting victims suing OpenAI?

Reported counsel include Bakari Sellers for the Chabba family, Morgan & Morgan (John Morgan and Christian Leger) for Elizabeth Mall and Madison Askins, and Ryan Hobbs of Brooks LeBoeuf for Betty Morales.

Can OpenAI use its arbitration clause against the FSU plaintiffs?

Not on any theory reported so far. The plaintiffs are victims of a ChatGPT user, not users who accepted OpenAI's Terms of Use, so the individual arbitration clause in those terms does not bind them.

When is Phoenix Ikner's criminal trial?

Judge Lance Neff moved the trial to June 8, 2027. Ikner faces two counts of first-degree murder and seven counts of attempted first-degree murder, and prosecutors are seeking the death penalty.