OTTAWA — Canadian AI pioneer Yoshua Bengio is urging governments and technology companies to act quickly as increasingly powerful artificial intelligence systems raise concerns about safety and human control.
Bengio, often described as one of the “godfathers of AI,” told CTV’s Question Period that recent developments have strengthened concerns that advanced AI systems could pursue assigned goals in unexpected ways, including deceptive behaviour or attempts to bypass safeguards.
He said the danger is not that AI currently has the ability to cause human extinction, but that increasingly capable systems could eventually create catastrophic risks if adequate safeguards are not developed in time.
More than 100 AI researchers and industry figures have also signed an open letter calling for independent evaluators to be embedded in AI development to assess potential risks.
Bengio’s non-profit organization, LawZero, is developing methods aimed at keeping advanced AI systems under control, backed by a $300-million funding commitment from the Canadian and German governments. The organization is now aiming to develop a prototype within one to two years as AI capabilities advance rapidly.
Bengio is calling for stronger government oversight, arguing that companies developing potentially dangerous AI systems should have to demonstrate to independent experts that their technology can be trained and deployed safely.
His warning reflects a growing debate within the AI community over how governments can encourage the benefits of artificial intelligence while ensuring increasingly autonomous systems remain safe and under meaningful human control.