Law360 (September 28, 2026, 2:04 PM EDT) -- OpenAI and its affiliates are urging a Florida federal court to throw out a suit from a woman alleging that its ChatGPT program encouraged the mass shooting at Florida State University in April 2025, saying it could not have known what the shooter planned and had no duty to warn in a way that could have prevented the shooting.

In a motion filed Friday, OpenAI said while the shooting was a "horrific act of violence," Vandana Joshi's complaint "stretches the bounds of state tort law beyond their breaking point."

According to the motion, Joshi's product liability claims fall flat because product liability law does not cover claims that target the communication of factual information, and the alleged shooter, Phoenix Ikner, only received from ChatGPT information that was already publicly available.

A product cannot be considered defective because of its content, OpenAI argued, particularly where the content consists of publicly available facts that could be found through any search engine.

In addition, OpenAI said the chat program is not a static, tangible product, but instead an "interactive and dynamic service," and the aspects of ChatGPT's service that are challenged here only concern factual information.

"No two ChatGPT users have the same expectations or usage; and even two users eliciting similar information may react and use that information in very different ways," OpenAI wrote. "Product-liability law is not designed to cover injuries allegedly resulting from a single person's idiosyncratic reactions to the information they receive."

In the suit, filed in May, Joshi, whose husband Tiru Chabba was killed, says that Ikner used ChatGPT to plan the April 2025 attack on the FSU student union in the months and days beforehand.

Ikner is criminally charged with killing two people and injuring six others in the attack, and is currently awaiting trial.

In Friday's motion, OpenAI also argued that product liability claims cannot extend to non-users of the service that are harmed by a user's intentional crimes or misuse of the service, adding that the chat logs made available to the court show that ChatGPT denounced violence and encouraged Ikner to seek professional help, but do not show that ChatGPT encouraged the violence.

That Ikner may have "viewed" ChatGPT's factual responses to his questions as encouragement does not amount to an allegation that OpenAI or ChatGPT "encouraged" the attack, the company wrote.

On Joshi's failure to warn claims, OpenAI argued that it had no duty to warn the general public, only its own customers and potential users, and the complaint does not explain how generalized warnings about the alleged risks of ChatGPT could have prevented Chabba's death.

In addition, the company wrote, the law does not require manufacturers to warn users of the potential consequences for criminal misuse of the products, and the complaint does not say how any additional warning to Ikner would have prevented the attack.

The negligence claims fail for similar reasons, OpenAI wrote, as Florida law does not recognize a duty to protect non-users against criminal violence, as OpenAI does not have a special relationship with the general public, nor did it have any right or ability to control Ikner, the weapons he used, or the place he used them.

And the allegations that ChatGPT or OpenAI failed to "connect the dots" or otherwise realize what Ikner was planning do not support a claim for negligence or that the company aided or facilitated the crime, the company wrote.

The company has no duty to "psychoanalyze" its users and warn the public about users that might be dangerous, OpenAI wrote, noting that Florida law does not even impose such a duty on medical professionals.

OpenAI also urged the court to throw out the negligent entrustment claims, saying that cause of action is meant for someone lending or allowing another to use equipment, such as a car or plane, in such a way that leads to injury or death, and does not apply to a chat program like ChatGPT, which is not in itself a "dangerous instrumentality." Nor could OpenAI have foreseen any risk of a shooting from Ikner's initial use of ChatGPT, the company added.

The company also argued that the First Amendment blocks the suit, as all of Joshi's claims boil down to holding OpenAI liable for disseminating lawful, publicly available information about matters of public concern, for which users may have legitimate uses.

The information did not incite Ikner to commit the shooting, nor can it be reasonably alleged to have aided and abetted Ikner in the crime, the company argued, and as such the information is not subject to any exceptions to the First Amendment's protections.

An attorney for Joshi declined to comment on Monday.

Representatives for the defendants could not immediately be reached for comment on Monday.

Joshi is represented by J. Robert Bell III and Gregorio A. Franci of Osborne Francis & Pettis, Bakari T. Sellers and Amy E. Willbanks of Strom Law Firm and James W. Bannister of Bannister Wyatt & Stalvey.

OpenAI is represented by Brian M. Willen, Samantha Machock and Matthew A. Macdonald of Wilson Sonsini Goodrich & Rosati PC and Douglas L. Kilby, Elizabeth Desloge Ellis and Hannah E. Murphy of Stearns Weaver Miller Weissler Alhadeff & Sitterson PA.

Counsel information for Ikner was not available on Monday.

The case is Joshi v. OpenAI, case number 4:26-cv-00222 in the U.S. District Court for the Northern District of California.

–Additional reporting by Emily Field. Editing by Vaqas Asghar.

For a reprint of this article, please contact reprints@law360.com.