OpenAI Considered Alerting Police Before Canadian Shooting Suspect’s Attack

Months before Jesse Van Rootselaar allegedly perpetrated a mass shooting in the rural British Columbia town of Tumbler Ridge, OpenAI employees identified troubling patterns in her interactions with ChatGPT and internally debated whether to alert law enforcement, according to multiple reports. The company’s abuse detection systems flagged her account in early 2025 for content suggesting violent intent, including detailed queries about firearms, tactical planning, and expressions of grievance targeting specific communities. Though internal discussions reached senior leadership, no formal report was filed with Canadian authorities before the tragedy unfolded.

According to The Wall Street Journal, OpenAI’s safety and abuse monitoring team raised alarms in June 2025 after automated systems detected a series of increasingly alarming prompts from the user associated with Van Rootselaar’s account. These included requests for step-by-step guidance on constructing explosive devices and discussions about targeting rural communities with minimal law enforcement presence. The company’s internal review noted that while the language did not constitute an immediate, verifiable threat under existing policies, the cumulative pattern met thresholds for heightened scrutiny. Nevertheless, OpenAI’s legal and compliance teams ultimately concluded that without corroborating evidence—such as a physical address, real-name identification, or direct threats to named individuals—they lacked sufficient grounds to notify law enforcement under current privacy and jurisdictional protocols.

MSNBC’s reporting corroborated that several OpenAI employees expressed frustration over the decision not to escalate the case. One anonymous safety engineer told investigators that the account’s behavior was "uniquely concerning," with patterns resembling those seen in prior mass shooting perpetrators studied by the company’s threat intelligence unit. The employee added that there was a "strong internal consensus" that the case warranted a proactive outreach to Canadian police, even if it meant navigating legal gray areas. However, company policy at the time required a higher burden of proof before initiating external notifications, particularly across international borders.

Yahoo News cited internal documents indicating that OpenAI’s abuse team had flagged Van Rootselaar’s account under the category of "furtherance of violent ideation," a classification reserved for users exhibiting sustained, detailed planning around acts of mass violence. The company’s policy explicitly prohibits the generation of content that facilitates illegal acts, and while ChatGPT repeatedly refused to provide actionable instructions, the user persisted in rephrasing queries to circumvent safeguards. OpenAI’s logs show over 80 interactions over a six-month period, with the user repeatedly asking for alternatives when denied information.

The shooting in Tumbler Ridge on October 12, 2025, left seven dead and several critically injured, triggering national mourning and renewed scrutiny of AI safety protocols. Canadian officials confirmed that Van Rootselaar acted alone and had no prior criminal record. Investigators have since requested access to OpenAI’s interaction logs as part of their forensic review. OpenAI has stated it is cooperating fully with authorities and has since revised its internal escalation protocols to allow for broader discretionary alerts in cases of high-risk ideation, even without concrete evidence of imminent harm.

This case has ignited a global debate on the ethical responsibilities of AI developers in preventing real-world violence. Critics argue that tech companies must prioritize public safety over legal caution, while defenders warn against preemptive surveillance that could chill legitimate discourse. As governments worldwide draft new AI regulations, the Van Rootselaar case may become a landmark reference point in determining where corporate duty ends and state responsibility begins.