Hinton Lectures to host Paul Christiano in Toronto Nov. 9

Paul Christiano, the AI safety researcher who helped develop reinforcement learning from human feedback at OpenAI, will give the third annual Hinton Lectures at the Metro Toronto Convention Centre from Nov. 9 to 11, the AI Safety Foundation said.

The three-night series runs in the John W. H. Bassett Theatre, is open to the public and will also be livestreamed. Tickets are available through hintonlectures.com, the foundation said in an Oct. 5 news release distributed by PR Newswire.

The lectures will open the inaugural AI Safety Week. The foundation, which runs the series, said its goal is to raise awareness and understanding of the catastrophic risks of artificial intelligence.

Christiano founded and leads the Alignment Research Center, a non-profit that studies how decisions are made inside machine learning systems. He is a senior technical adviser at the U.S. National Institute of Standards and Technology’s Center for AI Standards and Innovation, and sits on the OpenAI Foundation board and OpenAI’s Safety and Security Committee, according to the release.

Sinead Bovell, a Canadian futurist and the founder of the tech education initiative WAYE, will moderate. Manulife is the presenting partner, the Global Risk Institute is the founding partner and TELUS is the evening sponsor.

Ankesh Chandaria, chief executive of the foundation, said Canada helped build the foundations of modern AI, and that the key question now is what AI systems should do and who gets to decide.

Christiano laid out his own view in an interview with BetaKit published Oct. 8. He called the current period “a particularly dangerous moment” and said there is a real chance that within roughly six to 18 months humanity could struggle to prevail if AI systems tried to escape or undermine human control, BetaKit reported.

He also told BetaKit that reinforcement learning from human feedback, the technique he helped introduce, will probably break down as systems scale. He put his own research group’s odds of success closer to 10 per cent than 50 per cent. Those are his assessments, and they are not shared by all AI researchers.

He did not comment on specific governments or companies, including OpenAI, BetaKit reported. He said standards for measuring AI performance should not be settled behind closed doors and should involve the global scientific community, including experts in Canada.

The series is co-founded by Geoffrey Hinton, the University of Toronto professor emeritus and Nobel laureate often called a godfather of deep learning. A nominating committee that includes Hinton, Beth Barnes of METR, Roger Grosse of the University of Toronto and Shane Legg of Google DeepMind chose this year’s speaker.

Previous lecturers were Jacob Steinhardt of the University of California, Berkeley, in 2024 and Owain Evans of Truthful AI in 2025.

The lectures arrive as Ottawa builds its own AI policy. Prime Minister Mark Carney named 13 people to a new national AI council this week, and the federal privacy commissioner and provincial regulators have already completed a joint investigation into OpenAI’s ChatGPT. Neither body has commented on the lectures.

If you find reporting like this useful, consider supporting Canada News Media, an independent Canadian newsroom with no corporate parent.

Sources: AI Safety Foundation news release via PR Newswire (Oct. 5, 2026); BetaKit (Oct. 8, 2026).