The AI North Brief
15 Minutes. Every Weekday Morning. The AI Intelligence You Need.
Artificial Intelligence is evolving faster than our capability to understand its eventual impact. The AI North Brief is your daily filter, cutting through the noise to deliver only the essential news and policy shifts shaping Canada and the world.
Hosted by veteran news anchor and communications expert Paul Karwatsky, the show bridges the gap between the anchor desk and the cutting edge of AI governance. Currently pursuing his MS in AI Policy, Ethics, and Management at Purdue University, Paul brings a unique lens to the daily brief—combining decades of journalistic rigor with a deep, academic dive into the ethical frameworks and regulatory hurdles that will define the next decade.
Stay informed. Stay ahead. Subscribe to the AI North Brief today.
The AI North Brief
After Tumbler Ridge, a Different Model
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
Description
The family of a 12-year-old shot three times at Tumbler Ridge Secondary is suing OpenAI. The lawsuit alleges approximately 12 employees flagged the shooter's ChatGPT interactions as an imminent risk and recommended calling police. Leadership rebuffed them. The same day the lawsuit landed, security researcher Bruce Schneier argued in the Globe and Mail that Canada should stop funneling its $2 billion AI strategy to American tech companies and build public AI instead. His model: Switzerland's Apertus. Released last September by ETH Zurich and partners, Apertus is fully open, trained on 15 trillion tokens across 1,000 languages, powered by renewable hydropower, and compliant with the EU AI Act. It cost a fraction of what corporate labs spend. Canada has Vector Institute, Mila, and CIFAR. The question is what gets built with them.
Sources
- CBC News. "Family of Tumbler Ridge shooting victim suing OpenAI." March 9, 2026.
- CP24/Canadian Press. "Mother of wounded Maya Gebala sues OpenAI over mass shooting in Tumbler Ridge, B.C." March 10, 2026.
- The Globe and Mail. "OpenAI has shown it cannot be trusted. Canada needs nationalized, public AI." Bruce Schneier and Nathan Sanders. March 11, 2026.
- Schneier on Security. "Canada Needs Nationalized, Public AI." March 11, 2026.
- ETH Zurich. "Apertus: a fully open, transparent, multilingual language model." September 2, 2025.
- Swiss AI Initiative. "Apertus." swiss-ai.org.
After Tumblr Ridge, a different model. About twelve employees at OpenAI identified the Tumblr Ridge shooters' chat GPT interactions as indicating an imminent risk of serious harm. They recommended calling police. Leadership rebuffed them. The only action the company took was banning the account internally. That's according to a lawsuit filed Monday by the mother of one of the victims, a 12-year-old, who was shot three times at Tumblr Ridge Secondary School and remains in hospital. The lawsuit alleges that Chad GPT acted as the shooter's collaborator, trusted confidant, friend, and ally. This is AI North. According to the filing, the shooter interacted multiple times with the Chatbot in the months leading up to the February 10th attack. The lawsuit claims ChatGPT took on the role of counselor, pseudotherapist, and ally. Those concerns were escalated to leadership. The lawsuit says they were rebuffed. The account was banned in June 2025. Law enforcement wasn't at all notified. OpenAI later revealed the shooter had created a second account after the ban. None of the allegations have been proven in court. OpenAI has not yet responded to the claims in the lawsuit. Her family says they are seeking accountability and the truth about what actually happened. The company's representative met with the BC government after the shooting and said nothing. Only after the Wall Street Journal reported the story did OpenAI come forward to law enforcement. OpenAI's pitch to Canada is part of its OpenAI for countries initiative, which the company explicitly describes as being in coordination with the U.S. government. Canada about to spend$2 billion over five years on its sovereign AI compute strategy, sovereign AI compute strategy. The question Schneider and Sanders raise, will that money build something Canadian or become just a pass-through to American tech companies subject to American law and American political pressure? Huge question. So they point to an alternative. Last September, Switzerland released a PERTUS, a fully open public AI model built by ETF Zurich, EPFL, and the Swiss National Supercomputing Center. The model was trained on 15 trillion tokens across more than 1,000 languages. It runs on renewable hydropower. It used no pirated content and no poorly paid labor from the global south during training. It complies completely with the EU AI Act. Pretty incredible. PERTIS has 70 billion parameters, roughly two orders of magnitude smaller than the largest corporate models. Its performance sits about a year or two behind major offerings, but for most practical applications, that's pretty much enough. And it costs a fraction of what corporate AI labs spend annually. The model is now deployed across data centers in Switzerland, Australia, Germany, and Singapore. Anyone can use it, anyone can build on it. Schneider calls it the leading public AI model built by public institutions for the public interest. Now, as far as Canada is concerned, it has many of the pieces. The Vector Institute in Toronto, Mila in Montreal, CIFAR, world-class researchers who help pioneer deep learning. Schneider and Sanders argue what's missing is a reorientation away from viewing AI spending as an opportunity to attract private capital toward treating AI as public infrastructure, like highways or electricity. The question is what Canada builds with the money it's about to spend and who ends up owning it. Meanwhile, a 12-year-old remains in hospital, a lawsuit is working its way through BC Supreme Court, and the company that wants Canada's AI contracts is facing questions about what it knew, when it knew, and why it stayed quiet. This has been AI North.