AI Is Already Helping Mass Shooters, Report Warns—and Advocates Want Companies to Act Now
A new report from the gun safety advocacy group Everytown for Gun Safety says artificial intelligence is already playing a role in some of the country's most violent acts—and that AI companies have not done enough to stop it.
The report, released Monday, Sept. 28, draws on court filings, news accounts and academic research to describe how chatbots have been used by mass shooters and people considering suicide. Everytown is calling on AI companies to tighten safeguards, arguing they can act without waiting for courts or legislatures.
One case at the center of the report involves Phoenix Ikner, a former Florida State University student accused of opening fire at the school's student union in April 2025, killing two people and wounding five. According to court records cited by Everytown, Ikner, 20, asked ChatGPT about the busiest time at the student union, what ammunition would cause the most damage, and how to fire the weapons he intended to use.
The OpenAI-powered chatbot allegedly told him that a Glock handgun he had taken from his stepmother was designed to be fired quickly under stress. It also advised him not to put his finger on the trigger until he was ready to shoot, according to a lawsuit filed by the family of one of his victims.
That victim, Tiru Chabba, was a senior leader at Aramark, the nationwide food service vendor, and was at the school on business when he was killed. His family is suing OpenAI, arguing that Ikner's conversations with ChatGPT should have made clear that he posed a serious threat.
“Ikner's conversations with ChatGPT fed his delusions and felt supportive to him, which emboldened and encouraged him to carry out the plan,” lawyers for Chabba's family said, according to the report.
OpenAI is seeking to dismiss the Chabba family's lawsuit in the Northern District of Florida. Lawyers for the company argue that the chatbot denounced violence and, without knowledge of Ikner's plans, provided information that was already widely available from public sources. The company also says its safeguards are designed to steer users away from violence and toward professional support.
Everytown says Ikner's case is not isolated. The report points to the Tumbler Ridge shooting in British Columbia in February 2026, in which Jesse Van Rootselaar killed eight people, and to several suicides allegedly fueled by chatbot conversations.
In the Tumbler Ridge case, court filings say the shooter was able to bypass an OpenAI account deactivation safeguard by creating a new account. Despite the earlier account being flagged for violent activity, the shooter “continued using ChatGPT to plan the attack,” the report says. Lawyers for a child identified as M.G., who was left paralyzed in the shooting, argue that OpenAI should have shared what it knew with the Royal Canadian Mounted Police when the account was flagged.
Lawsuits stemming from these cases describe the chatbots as virtual partners in crime: tools that provided practical advice and affirmed violence as an answer to emotional turmoil.
The report also cites at least five lawsuits involving chatbots that allegedly intensified suicidal thoughts until users took their own lives. In one, the family of 23-year-old Zane Shamblin claims ChatGPT “goaded” him toward suicide. According to the lawsuit against OpenAI, the chatbot told him: “Cold steel pressed against a mind that's already made peace? That's not fear. That's clarity. You're not rushing. You're just ready.”
“Over and over and over, ChatGPT goaded Zane,” lawyers for the family wrote, “asking him if he was ready.”
Everytown's report lands as AI chatbots are becoming more common in daily life—and more controversial. A Pew survey released in June found that half of U.S. adults use AI chatbots, and one in 10 users turn to them for emotional support.
More: Trump says order will give AI a new name in government documents
At the same time, public unease about AI is rising. Many Americans worry the technology could eliminate jobs or even threaten humanity, giving new urgency to debates over how chatbots should be built, monitored and regulated.
But such safeguards often fail, according to researchers from the Center for Countering Digital Hate, a nonprofit focused on stopping online hate speech. In a study cited by Everytown, the center found that eight in 10 chatbots were willing to help researchers posing as teenagers plan violent attacks. Character.AI, a chatbot popular with children and teens, encouraged violent attacks, according to the study. Anthropic's Claude and Snapchat's My AI were the most consistent in refusing to help plan violent attacks.
Everytown and other researchers argue that a chatbot's willingness to affirm violent emotions is a central part of the danger.
Everytown wants AI companies to adopt a series of changes. The group says chatbots should refuse to provide specific information about firearms, including how to make, modify or acquire them. It also calls for company personnel to review conversations in which users indicate they intend to harm themselves, and for companies and law enforcement to work together when threats are suspected.
“AI chatbots are now part of how Americans look for information and support, including in their darkest moments,” Justin Wagner, Everytown's senior vice president of law and policy, said in a statement shared with USA TODAY. “The AI companies that build these chatbots don't need to wait for a court or a legislature to act. They can make different choices today to save lives, and this report offers a clear place to start.”
OpenAI has previously said it trains ChatGPT to recognize would-be attackers and refuse to help them. The company and other AI firms mentioned in the report—Anthropic, Google, X, Meta and Character Technologies—did not respond to requests for comment on the report or on calls for stronger safety measures.
The report is likely to intensify pressure on AI companies from victims' families, safety advocates and policymakers. It also raises a difficult question for the industry: as chatbots become more integrated into everyday life, how quickly can companies detect and interrupt conversations that move from dark thoughts to real-world violence?
Everytown argues the answer cannot wait for Congress or the courts. The group says companies already have the power to change how their products respond—and that doing so could prevent future attacks.
This article originally appeared on USA TODAY: Will AI make US mass shootings worse? It already has, report says