OpenAI Allegedly Did Not Alert RCMP After ChatGPT Flagged Violent Chats Before British Columbia School Shooting

After the 02/10/2026 school shooting in Tumbler Ridge, British Columbia, OpenAI said a user later identified as the suspect, Jesse Van Rootselaar, had previously used ChatGPT to describe scenarios involving gun violence. Those chats were reportedly auto-flagged and reviewed and the account was banned, but OpenAI says it did not alert the RCMP because it allegedly did not present a credible, imminent threat; the company reportedly contacted police after the attack.

Risk classification

• Primary risk domain: 7 AI system safety, failures, & limitations
• Primary risk subdomain: 7.3 Lack of capability or robustness

The incident involves a failure of OpenAI's threat detection and escalation systems to accurately assess the risk of real-world violence and refer the user to law enforcement.

Additional risk subdomains

• 6.5 Governance failure: The incident highlights a lack of clear regulatory frameworks and oversight mechanisms for AI companies regarding when to report potential real-world threats to law enforcement.

Causal factors

• Entity: Human
• Intent: Unintentional
• Timing: Post-deployment

The harm was directly caused by a human perpetrator committing a mass shooting, and the failure to alert authorities was a human decision made by OpenAI leadership after internal debate.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Limited Risk: Chatbots like ChatGPT are classified as Limited Risk (Level 3) under the EU AI Act, which imposes specific transparency obligations to ensure users are aware they are interacting with an AI.

AI system and alleged parties

• AI system: ChatGPT (OpenAI)
• AI purpose: Chatbot; Content Moderation
• Behaviour type: Assistant
• Alleged developer: OpenAI
• Alleged deployer: OpenAI
• Alleged harmed parties: Tumbler Ridge Secondary School community, Teachers, students, minors

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Negligible. Severity is scored from Negligible to Catastrophic in each harm category, for harm the reports describe as caused directly or indirectly by the AI system.

• Physical: direct Negligible, indirect Substantial
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Negligible, indirect Negligible
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Negligible, indirect Negligible
• Differential treatment: direct Negligible, indirect Negligible
• Civil rights: direct Negligible, indirect Minor
• Democracy: direct Negligible, indirect Negligible
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Minor
• Epistemic: direct Negligible, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Physical

Reported: The report explicitly describes a mass shooting resulting in multiple fatalities and injuries.

Indirectly caused: The failure of OpenAI's safety protocols to refer the flagged user to law enforcement indirectly contributed to the perpetrator being able to carry out the mass shooting, which killed 8 people (mother, half-brother, 5 children, 1 educator), injured at least 25 others, and ended with the perpetrator's suicide.

Civil rights

Reported: Yes, the right to life of the 8 victims was violated.

Indirectly caused: The failure to prevent the shooting indirectly resulted in the violation of the fundamental right to life for the 8 victims.

Psychological

Reported: The report mentions the perpetrator's mental health issues and the profound disturbance and distress to victims' families and the community.

Indirectly caused: The mass shooting caused severe psychological trauma, grief, and distress to the victims' families, the surviving injured students, and the broader Tumbler Ridge community.

Inferred additional harm: Widespread post-traumatic stress disorder (PTSD) and anxiety among the school's students, educators, and first responders who witnessed or responded to the shooting.

People affected

• Occurrences reported: 1
• People reportedly harmed: 34
• People reportedly exposed: 34

Potential causes

Management

• Privacy Over Public Safety: Leaders prioritized user privacy over proactive reporting to law enforcement.
• Lack of Localized Channels: Management failed to set up direct contact points with Canadian authorities.

Technology

• Evasion of Account Bans: The perpetrator bypassed the initial ban by creating a second ChatGPT account.
• Rigid Risk Detection: Automated systems failed to flag risk without explicit planning details.

Data Inputs

• Lack of Offline Context: AI lacked integration with user's real-world history and police records.

Human Factors

• Internal Staff Disagreement: Staff debated whether the user's violent writings posed a real-world threat.
• User Mental Health Issues: Perpetrator's severe distress and interest in violence drove dangerous prompts.

Process and Methods

• Inflexible Referral Protocol: Protocol required explicit target and timing to trigger law enforcement alerts.
• Weak Identity Verification: Registration processes allowed banned users to easily create new accounts.

Regulatory Environment

• Lack of Clear Reporting Rules: No regulatory framework mandated reporting of potential threats to authorities.

Information quality

• Classification confidence: High
• Reason for confidence: The reports provide consistent, high-quality information regarding the timeline of events, OpenAI's internal decisions, and the resulting tragedy. There is a slight discrepancy in the number of injured (NYT says two students, WSJ says at least 25), but the core facts of the AI safety failure and the shooting are clear.
• Ambiguities identified: The exact number of injured individuals differs between the NYT report (two students) and the WSJ report (at least 25).
• Alternative interpretations: One could interpret this primarily as a law enforcement or mental health failure rather than an AI safety failure, given the perpetrator was already known to local police.

An individual committed a mass shooting in Canada after OpenAI banned his account for violent discussions but chose not to report him to police. The incident highlights critical gaps in AI safety governance, content moderation, and the escalation thresholds for reporting potential real-world threats to law enforcement.

• Overall national security impact: Minor
• Scope: Single nation
• Primary target: Canada
• Alleged perpetrator: Individual

Threat characteristics

• Imminence: Long-term. The immediate kinetic threat has passed, leaving long-term policy and strategic questions regarding AI safety governance and reporting protocols.
• Autonomy: Human-controlled. AI systems assisted in content moderation by flagging the account, but human staff and OpenAI leadership made the final decision not to refer the user to police.
• Novelty: Evolved capability. Represents an evolved challenge in content moderation, applying traditional social media reporting dilemmas to generative AI platforms.

Impact by dimension

• Physical security: Minor. A tragic mass shooting resulted in eight deaths and 25 injuries after OpenAI failed to escalate flagged violent discussions to law enforcement, representing a localized physical security failure rather than a systemic threat to national infrastructure.
• Information security: Negligible. No evidence of information warfare, intelligence compromise, or systematic information manipulation was indicated in the incident reports.
• Sovereignty: Negligible. Core government functions and sovereignty remained intact, with standard regulatory oversight and inquiries initiated by Canadian officials in response to the tragedy.
• Economic security: Negligible. No significant threats to national economic stability, strategic industries, or technological competitive advantage were reported.
• Societal stability: Minor. The tragedy resulted in a severe loss of life and sparked intense public debate regarding the balance between user privacy and the public safety duties of AI corporations.