Canada calls on OpenAI for safety measures after ChatGPT misuse

Canada has called on OpenAI to submit a detailed safety plan following its decision not to alert police about a teenager’s violent use of ChatGPT, according to officials.

The government’s AI minister, Evan Solomon, met with OpenAI leaders in Ottawa after the company disclosed that an 18-year-old suspect, Jesse Van Rootselaar, was flagged for misuse in June 2025 but was not reported to authorities.

Van Rootselaar is accused of killing six children and two adults in Tumbler Ridge, British Columbia, and is believed to have died by suicide.

OpenAI said it is cooperating with law enforcement and plans to provide more safety proposals tailored to Canada.

The government is also reviewing measures to establish clearer standards and accountability for AI systems used in the country.

The meeting involved officials from public safety, justice, and culture, and discussed how to improve risk detection and response, especially involving minors.

🔗 Source: Bloomberg

🧠 Food for thought

Implications, context, and why it matters.

OpenAI split internally over whether to report the shooter

OpenAI chose not to contact police after an internal review, rather than missing a step.
The suspect’s ChatGPT account was flagged and banned in June 2025, months before the attack, for writing scenarios involving gun violence 1.
About a dozen employees argued over next steps, with some pushing leaders to contact Canadian law enforcement 1.
Leaders declined, saying the material fell short of the company rule for an imminent and credible risk of serious physical harm to others, which triggers escalation 1.

The incident raises a hard global debate on AI and public safety

The Tumbler Ridge tragedy moves AI safety from abstract risk into a live policy problem for governments worldwide.
It adds pressure on Canadian lawmakers to set a clear and transparent threshold for when tech firms must report threats, a request from British Columbia’s premier 2.
Writing those rules stays difficult, since experts warn mandates to report every possible suspicion could flood police with low-grade tips 2.
The case may speed up online harms legislation in Canada and act as a reference point for countries weighing privacy against public safety 2.