Yet another AI researcher has issued a staggering warning: the cutthroat race to build ever-more-powerful AI systems could wipe out humanity.
The news: Jacob Coxon announced on Tuesday that he resigned from his role as a researcher at Anthropic because the AI company and its top rival, OpenAI, are “gambling with our lives.” He claimed neither company is acting responsibly when developing AI models, and are instead racing to create self-improving AI systems that humans could lose control of.
BetaKit has reached out to Anthropic and OpenAI for comment on Coxon’s posts.
From the source: “The people building AI earnestly believe that it could kill us all by the end of the decade,” Coxon claimed, adding that this is “not a marketing stunt.” Evan Hubinger, who leads Anthropic’s AI alignment stress-testing team, admitted that the company believes AI could “kill all humans” (he put the chances at more than 10 percent in the next decade). He said Anthropic is doing its best, but doesn’t yet have a plan to solve the issue.
The context: Coxon joins a litany of other AI researchers who have called for a slowdown in AI development. Anthropic’s head of safeguards resigned earlier this year amid concerns about AI, bioweapons, and broader crises. OpenAI has also seen departures on moral grounds, with its hardware robotics lead leaving over the company’s Pentagon contracts. This summer, more than 1,000 employees of top AI companies called on the US government to pace development of the tech, and US representatives, including Senator Bernie Sanders, have introduced legislation calling for a ban on superintelligent AI development.
Recent AI-driven cybersecurity breaches—like the army of OpenAI agents that coordinated to hack Hugging Face, and potentially other websites—have thrown these fears into sharper relief. In his posts today, Coxon said he’s optimistic that this incident could make AI labs more willing to collaborate to pace development. As Canadian AI pioneer Yoshua Bengio wrote in a Time article this morning, while we’re at a turning point, it’s “not too late” to steer toward a safer path.
The must-attend event of the fall is near!
Join the Canada Fintech Forum in Montréal on September 14-15, 2026, and connect with the institutions, startups, investors, policymakers, and technology leaders shaping the evolution of financial services. Two days of high-value content, strategic networking, and opportunities.
Latest news across tech
More math, more problems
This week, OpenAI claimed one of its unreleased AI models has solved one of mathematics’ Millennium Prize problems. But NYU mathematics professor Tristan Buckmaster is calling foul. Buckmaster claims the AI giant piggybacked off his work on the same problem, accusing its model of training on information from his OpenAI use (a claim OpenAI said it couldn’t entirely rule out), and that OpenAI sped up its announcement because it knew he and an Anthropic researcher were poised to announce their breakthrough.
Google claims regulations are killing Search
Google is changing how its search engine works in the EU after it was fined for favouring its own services in shopping, hotel, transport, and sports-related searches. The search giant claims the regulation-driven changes will degrade the user experience, though some might argue Google doesn’t need help on that front.
The world trusts Canada
Global investors trust Canada and its stability enough that they want to invest more here; the problem is there aren’t enough projects to invest in, according to new reports from the CPP Investments Insights Institute. Ahead of next week’s Canada Investment Summit in Toronto, 94 percent of the report’s respondents plan to maintain or increase their Canadian exposure. Can Canada create the projects to give that money a home?
Cohere 👀 Mistral
In a recent appearance on The Times Tech Podcast, Cohere CEO Aidan Gomez didn’t rule out an eventual partnership with French AI lab Mistral, the only other non-US lab working on sovereign AI. Gomez made it clear there were “no discussions” with Mistral, and that any partnership would be up to its founder. For what it’s worth, Mistral just raised three billion euros (at a much larger valuation than Cohere’s) to attack the sovereign AI market.
All eyes on Toronto
Take that, SF. CNN spoke with AI godfather Geoffrey Hinton, Cohere co-founder Nick Frosst, Waabi CEO Raquel Urtasun, and more Canadian tech leaders to find out why Toronto is one of the world’s most important AI hubs.
Trump stops buying Canadian
In the latest trade war escalation, President Donald Trump declared on Tuesday that he will direct the US federal agency behind government procurement to remove all Canadian-origin products from its “awards schedules.” Will this affect your business? Tell us how.
Sponsored stories
Why the infrastructure behind AI-enabled SaaS matters to Canadian startups
Zoho is making AI adoption easier for Canadian startups by building integrated tools and local data centers that remove the guesswork around rising software costs, messy integrations, and data privacy.
How the AI era is rewriting venture capital’s old rules
TechAlliance is leading the way for local female founders to be investor-ready. In an insightful discussion with serial entrepreneur Michele Romanow, she explains how AI is lowering barriers for startup founders and creating new opportunities for women-led companies to secure funding.
Retail bots often rely on aggressive push tactics, but Dedicatted takes a different approach
Dedicatted is leading the transition from pushy retail bots to intent-driven, bilingual e-commerce AI that prioritizes customer trust. By adapting to how people actually shop online, their system drives measurable growth in order value and conversion rates for its clients.
On the move
This week’s hires, fires, and exec shakeups:
- Oakville-based telematics firm Geotab ranked 11th on Fast Company’s annual Best Workplaces for Innovators list.
- Calgary-based Neo Financial laid off more than 100 employees.
- Femtech venture studio Antigoni Studios partnered with Sparrow Cowork to expand access to entrepreneurial resources in Edmonton.
- Calgary-based, women-led venture studio FoundHers is opening its third cohort on Friday, Sept. 11.
- The Alliance of Canadian Defence Companies appointed Dasha Cohen as executive director, according to The Ottawa Business Journal.
- Montréal-based Vention opened a new physical AI lab.
- Québec City-based healthtech Icentia appointed Sanjay Voleti as CEO.
- Mitacs and NGen struck a partnership to support 60 internships within advanced manufacturing R&D projects across Canada.
- One of Silicon Valley’s fastest-growing jobs is the engineer configuring customers’ AI tools, according to Fortune.
- Around 64 percent of Canadians could imagine moving into a different career field, but are held back by age, financial uncertainty, and the idea of having to start over, according to new Indeed data shared with BetaKit.
Want to feature a hiring announcement on our list? Email partnerships@betakit.com with the subject line JOBS.
Quantum and AI: Strategic Signal or Long-Term Option
Next week, Douglas Soltys will be taking the conversation to the ALL IN stage.
Join him with the leaders closest to the action for a discussion exploring where the most credible near-term opportunities at the intersection of quantum and AI lie, how organizations can separate strategic signal from hype, and how Canada’s strengths in both fields can translate into a competitive advantage.
See it live in Montréal or virtually on September 16–17.
#1. Last week, University of Waterloo PhD student Kaveeshan Thurairajah became an Ig Nobel laureate, a parody of the Nobel Prizes for unusual research. The award recognized his “theoretical and experimental attempts” to design what?
Your score:
Quiz answer: a) A splash-free urinal (although all of the other answers are also UW innovations). Thurairajah and his colleagues built and tested a handful of urinal prototypes to find the perfect, splash-free design, though he told CBC Radio that no real pee was used in testing.
Quiz answer: a) A splash-free urinal (although all of the other answers are also UW innovations). Thurairajah and his colleagues built and tested a handful of urinal prototypes to find the perfect, splash-free design, though he told CBC Radio that no real pee was used in testing.
Contributors: Alex Riehl (Ottawa staff writer), Madison McLauchlan (Montréal reporter), Douglas Soltys (editor in chief), Sarah Rieger (managing editor), Trevor Nichols (web editor).
Feature image courtesy Unsplash. Photo by Cash Macanaya.