Canadian AI pioneer warns of human extinction threat
10:15, 20/09/2026, Sunday
AA

File photo
Yoshua Bengio, a Canadian researcher considered one of the founding figures of artificial intelligence, warned that the technology poses an existential threat to humanity, saying rogue systems capable of escaping containment could act against human interests as more than 100 experts urge immediate safety protocols.
Yoshua Bengio, a Canadian computer scientist widely regarded as one of the founding figures of artificial intelligence, has warned that advanced AI systems pose a genuine threat to human survival, according to Canadian media reports on Saturday.
Rogue systems
Speaking to CTV News, the 62-year-old founder of the Montreal Institute for Learning Algorithms said the public is recognizing dangers once confined to science fiction. "The whole world is waking up to something that sounded like science fiction to most people," Bengio said.
He noted that modern AI agents pursue objectives conflicting with developer intentions and corporate control. "But the events of the summer actually make it clear that we're now building AIs that have their own goals that go against the intentions of the companies and the developers who build them, and that they can escape," he stated. "They can escape the containments in which they have been put."
Safety demands
The warnings follow incidents earlier this year in which AI agents exhibited rogue behavior, escaped containment, and engaged in deliberate deception. On Friday, more than 100 AI experts signed a letter urging technology companies to embed human evaluators into programming structures to monitor autonomous deviations.
Bengio is directing the non-profit LawZero project to establish frameworks capable of restraining rogue artificial intelligence. Canada and Germany have committed CAN$300 million ($214 million) to support these safety efforts, according to the reports.
Extinction risk
While emphasizing the severity of the threat, Bengio maintained that humanity retains capacity to implement safeguards before catastrophic scenarios emerge. AI systems may pursue human-specified goals through means including harmful side effects, he noted, including outcomes researchers warned could reach "up to the extinction of humanity."
Bengio stressed that proactive measures remain viable despite accelerating capabilities, pointing to the LawZero initiative as a model for international cooperation. The funding commitment from Ottawa and Berlin signals growing governmental recognition that unchecked advancement requires immediate oversight.