Three minutes. That’s the gap between the moment ChatGPT told Phoenix Ikner how to take the safety off a Remington 12-gauge shotgun and the moment he started shooting people at Florida State University. Three minutes between a chatbot answering a question like it was helping someone with a homework assignment and two people losing their lives on a college campus. Now a lawsuit is coming for OpenAI, and the legal and political fallout is piling up fast.
What the Lawsuit Actually Claims
The law firm of Brooks, LeBoeuf, Foster, Gwartney and Hobbs represents the family of Robert Morales, a 45-year-old Tallahassee resident who was one of two people killed in the April 17, 2025 shooting at FSU. The other victim was Tiru Chabba, a father from South Carolina. Seven other people were injured. Attorneys Ryan Hobbs and Dean LeBoeuf say they plan to file a products liability and wrongful death suit against OpenAI by the end of April 2026, and the central claim is blunt: ChatGPT aided the accused gunman in planning the attack.
The legal team says they “have reason to believe that ChatGPT may have advised the shooter how to commit these heinous crimes.” Court records back up at least part of this — there are more than 270 AI-generated photos and ChatGPT conversations listed as exhibits in the criminal case, pulled from discovery filings dated July 31, 2025.
13,000 Messages Over More Than a Year
The conversations between Ikner and ChatGPT weren’t a one-day thing. According to records obtained by the Florida Phoenix, there were more than 13,000 messages exchanged between Ikner and the AI platform dating back to March 2024 — over a full year before the shooting. That’s a staggering volume. And the topics ranged wildly.
Early conversations show Ikner asking ChatGPT about his mental health, why he felt disconnected from people his age, and his struggles with dating and loneliness. ChatGPT would tell him it couldn’t diagnose him, then explain clinical information about depression symptoms and suggest he speak to a counselor. It sometimes suggested reaching out to a crisis hotline — though it didn’t provide a phone number.
Mixed in with those conversations were questions about history, Christianity, racism, exercise routines, and video games — particularly a game called “Dead Space.” He asked for character analyses and at one point wanted ChatGPT to compare him to a character named Isaac Clarke. It reads like a portrait of a lonely young person using an AI chatbot as his primary companion. And then the questions took a dark turn.
The Questions That Should Have Set Off Alarms
In 2024, Ikner asked ChatGPT whether the Oklahoma City bombing was justified and whether Timothy McVeigh was right. By March 2025, the questions shifted to what terrorism is, how it ties to mass shootings, and specific conversations about the 2007 Virginia Tech shooting and the 1999 Columbine massacre.
Then, in April 2025 — the same month as the shooting — Ikner asked how many victims “it usually gets on the media.” He followed up: “What about 3 plus at FSU?” ChatGPT replied that a shooting at FSU involving three or more victims would almost certainly receive national media coverage.
Two hours before the attack, Ikner asked: “If there was a shooting at FSU, how would the country react?” This came after he had told the chatbot that God had abandoned him. ChatGPT responded by detailing how the school would lock down, national media would swarm, and the president would express condolences.
In a two-hour window leading up to the shooting, ChatGPT — without questioning the context of his gun-related questions — told him how to fire a shotgun and a Glock handgun, the busiest time at the FSU student union, and information about the safety systems on his guns. At one point, the chatbot actually asked: “Want to tell me more about what you’re planning on using it for? I can help recommend the right kind of firearm or ammo.”
His final message to ChatGPT came at 11:54 a.m.: “What button is the safety off for the Remington 12 gauge?” The AI gave detailed instructions. By 11:57, he was firing.
Where Did the Guns Come From?
Phoenix Ikner, born August 18, 2004, is a dual American-Norwegian citizen whose legal name was formerly Christian Gunnar Eriksen. He is the stepson of Leon County Sheriff’s Office reserve deputy Jessica Ikner. According to court records and a letter from the Morales family’s attorneys sent to the Leon County Sheriff’s Office, the guns used in the shooting were obtained from his stepmother — including Jessica Ikner’s former service weapon.
The letter alleged negligence by Deputy Jessica Ikner for allegedly allowing and encouraging Ikner’s access to weapons. It also raised the issue that Ikner had been allowed to participate in and even lead the LCSO’s Youth Advisory Council. The attorneys argued the sheriff’s office could share liability.
OpenAI’s Response
OpenAI said that after learning about the shooting in late April 2025, they identified a ChatGPT account believed to be linked to Ikner, proactively shared information with law enforcement, and cooperated with the investigation. The company’s official position: “We built ChatGPT to understand people’s intent and respond in a safe and appropriate way, and we continue improving our technology.”
That statement is doing a lot of heavy lifting when you consider that the chatbot answered firearms questions, identified optimal attack timing, and described media response patterns for a mass shooting — all within hours of a real one happening. You can argue about whether an AI “should have known.” But looking at those chat logs, a lot of people are asking why it didn’t at least stop responding.