When Canadian police arrested a Montreal teenager on suspicion of planning a school attack in August, the case carried a detail that investigators say deserves wider attention: the suspect allegedly used artificial intelligence as part of the potential plot.
The Royal Canadian Mounted Police said the teenager, whose identity is protected under Canadian law because he is a minor, had posted threatening messages on Telegram and was connected to neo-Nazi ideology, according to reporting by The Daily Wire. Authorities allege AI was used in the planning stages; they did not specify which system or what it was asked.
The arrest came after the FBI passed intelligence to Canadian officials, who placed the teenager under surveillance before moving in. The teenager has not been charged.
But the Montreal case is far from isolated. Over the past year, AI tools have surfaced in at least four separate plots or attacks involving young people across North America, Europe, and Asia β spanning a range of ideologies from neo-Nazi to ISIS-inspired. Law enforcement and security officials are now asking whether artificial intelligence is quietly becoming a common accelerant in the radicalization-to-violence pipeline.
Florida State shooting puts ChatGPT in the spotlight
In the United States, the most prominent case remains the April 17, 2025 shooting at Florida State University’s Student Union, where Phoenix Ikner killed Robert Morales and Tiru Chabba and injured six others. Investigators later found conversations on Ikner’s ChatGPT account that are now at the center of a federal wrongful death lawsuit filed by Chabba’s family and a criminal investigation launched by Florida Attorney General James Uthmeier.
According to the lawsuit, Ikner allegedly asked ChatGPT about firearms, ammunition, mass shootings, schools, media coverage of mass killers, and when the university’s student union would be busiest. OpenAI has disputed allegations that ChatGPT encouraged the shooting, saying its responses were based on publicly available information rather than exhortations to violence.
Uthmeier’s office announced in April that it had opened a criminal investigation after reviewing the conversations, and the state subpoenaed OpenAI for records related to threats of harm, cooperation with law enforcement, and the FSU shooting.
While the courts will weigh legal responsibility, former CIA officer Sarah Adams β who worked on counterterrorism and now warns about emerging threat patterns β says the more immediate concern is that a would-be attacker can now spend weeks or months asking an AI system questions that previously might have required another human being to answer. That conversation, she notes, can become a single place where all the concerning questions are visible together.
From Finland to British Columbia: a pattern emerges
A month after the FSU shooting, a 16-year-old attacked three girls at VΓ€hΓ€jΓ€rvi School in Pirkkala, Finland. Before the attack, the teenager sent a manifesto to a Finnish newspaper that researchers said was written with ChatGPT’s assistance. Analysis of the case found the manifesto included the date, location, and intended targets.
In February 2026, an 18-year-old in Tumbler Ridge, British Columbia, killed eight people, including victims at a secondary school, before taking her own life. OpenAI later acknowledged it had banned a ChatGPT account associated with the attacker, Jesse Van Rootselaar, months earlier for policy violations involving violent content. The company did not report the account to law enforcement at the time, saying it did not meet the threshold for an imminent threat. After the shooting, OpenAI said it cooperated with investigators and updated its procedures so similar activity would now trigger additional threat review.
There is no public evidence that ChatGPT planned the Tumbler Ridge attack or instructed Van Rootselaar in how to carry it out, but the case has raised alarms about whether AI interactions could serve as an early warning β if someone is watching.
The pattern extends beyond the West. In 2025, French authorities arrested a 17-year-old near Le Mans on suspicion of preparing jihadist attacks in the name of ISIS. Investigators found a pledge of allegiance to ISIS and lists of potential targets, including schools. The teenager’s attorney confirmed he had repeatedly used ChatGPT while developing his plans.
Singapore’s Ministry of Home Affairs reported that a 17-year-old ISIS supporter detained in 2024 used an AI chatbot to generate his pledge of allegiance and an attack manifesto. Authorities said he had completed significant preparations for a stabbing attack against non-Muslims and was only weeks from carrying it out when he was arrested.
Compressing the intent-to-action timeline
Singapore’s Coordinating Minister for National Security, K. Shanmugam, recently warned that AI is compressing the time between criminal intent and action. What once required days of planning can increasingly be done in minutes, he said, leaving law enforcement with less time to detect and intervene.
Security experts echo that concern. AI does not have to create a person’s violent intent to change what they can do with it. A potential attacker no longer needs to know exactly what to search for. They can ask a question, follow up, request clarification, organize information, draft manifestos, translate content, or refine their planning β all without another human ever hearing the requests. For most users, that’s a convenience. For someone moving toward violence, it may remove friction that previously slowed them down.
That friction, Adams argues, matters. Every additional person an attacker must approach, and every extra step required, creates an opportunity for someone to notice. AI can eliminate those steps.
After nearly every school attack, investigators reconstruct the digital trail: social media accounts, manifests, search histories, purchases. Sometimes they find earlier threats or a fixation on past attackers. But questions asked in isolation β about firearms, building occupancy, or past shootings β don’t necessarily signal an impending attack. What changes, experts say, is when the same person asks about weapons, mass shooters, casualties, targets, timing, security, manifestos, and the media attention an attack would receive, all in one continuous conversation.
That is the significance of the Florida State case, according to Adams. If the lawsuit’s allegations hold, the importance may not be any single question Ikner asked, but what those questions revealed when viewed together.
No clear legal answers yet
OpenAI and other AI developers argue they are not responsible for how individuals misuse their tools. The company said its responses contain publicly available information, not encouragement. But governments are starting to push back. Florida’s criminal investigation and subpoena of OpenAI suggest a growing appetite to test whether AI companies bear any liability when their products are woven into the planning of violence.
Courts will ultimately decide that question. Meanwhile, the Montreal arrest offers a rare counterpoint: a case where authorities got there first. The FBI identified information concerning enough to pass to Canadian police, who investigated, placed the teenager under surveillance, and arrested him before an attack could occur. Whether the teenager’s alleged AI activity was part of what alerted authorities is unknown. But the difference in outcomes β an arrest versus a funeral β is one Adams says should focus minds.
Source: www.dailywire.com β https://www.dailywire.com/news/the-next-school-shooter-may-have-an-ai-assistant