On February 10, an 18-year-old girl shot and killed her mother and brother. Then she went to the institute and murdered seven more people before committing suicide. Disturbingly, the shooter had discussed her violent intentions with ChatGPT, and OpenAI had the opportunity to notify the police but chose not to.
What Has Happened?
As reported by the Wall Street Journal, OpenAI’s automated system detected messages from a user describing scenarios of armed violence as early as June of last year. Some employees raised alarms, emphasizing that these communications could lead to real violence, sparking internal debates about whether to alert Canadian authorities. Ultimately, they closed the user’s account but did not follow through with any notifications. Now, Canadian authorities have summoned them for clarification.
More Alarming Cases
The Tumbler Ridge shooting is not an isolated incident. In early 2025, a man parked a Cybertruck full of explosives in front of a Las Vegas hotel with the intent to detonate it (the only victim was himself). Prior to this, he consulted ChatGPT for guidance on execution. Notably, the chatbot failed to flag any concerning messages, a fact only discovered retrospectively when OpenAI reviewed the interactions.
In another case in Seoul, a woman was incarcerated for allegedly murdering two individuals through benzodiazepine poisoning. Investigations revealed that she had consulted ChatGPT to determine lethal dosages and effects when mixed with alcohol. While her inquiries might stem from genuine curiosity, this serves as another example of how AI technology can be involved in criminal activity.
Why It Matters
Artificial intelligence has transformed into a digital confessional, a space where individuals share secrets, including their darkest thoughts. Some treat AI like a friend or confidant. Given this context, it shouldn’t be surprising that someone might relay intentions of violence to a chatbot. The real concern lies in what companies are doing to mitigate these risks, which currently appears insufficient.
Ethical Obligation vs. Legal Requirement
When therapy sessions become confessions of harmful intentions, professionals are required to contact authorities—a practice grounded in ethical standards. However, no legal framework obliges AI companies to disclose violent intentions revealed by users, leaving it to their discretion. The onus of reporting becomes one of ethics rather than law.
Homemade Bomb Instructions and Internet Precedence
Incidents like the Tumbler Ridge shooting aren’t solely a product of AI chatbots. Manuals for creating homemade explosives have circulated for decades, raising concerns well before the Internet became widely accessible. Additionally, information on suicide is readily available online, negating the necessity to consult AI platforms like ChatGPT for such matters.
The Complicated Nature of AI Conversations
According to a former OpenAI employee, engaging with a chatbot often reveals clearer intentions through extended dialogue, unlike a simple search. This capability could enable easier detection of harmful intentions like those displayed by the Tumbler Ridge shooter. However, the risk of false positives escalates, especially when users might be crafting fictional narratives or role-playing scenarios. Navigating this complexity remains a significant challenge for AI companies.
In summary, while the potential for AI to assist in criminal planning is undeniable, the ethical and legal responsibilities of AI companies remain largely unaddressed. As technology continues to evolve, so too must our approach to safeguarding society against its misuse.