Paul Christiano has warned there is a “high risk” of an AI takeover within the next three years and even a “meaningful risk” within six to 12 months.
The Alignment Research Center founder, 33, who also serves on the OpenAI Foundation Board, delivered the chilling forecast Monday at the AI Safety Foundation’s 2026 Hinton Lectures in Toronto, Canada. Christiano told attendees during his opening address that artificial intelligence poses the most probable cause of death for many in the room.
“I think for most people in this room, the most likely reason they will die is because of an AI accident,” Christiano said during his opening address on Monday.
Nobel Prize-winning computer scientist Professor Geoffrey Hinton, 78, widely dubbed the “Godfather of AI,” shared the stage and admitted his own timeline estimates now look conservative. Hinton, who informally advises both the US and UK government on artificial intelligence, previously believed existential AI risks were at least a decade away.
“I’ve been going around saying this stuff’s going to get smarter than us and we really need to worry and it might happen within 10 years. And so I’m the one who is scaring other people,” Hinton said on Wednesday. “But I go to Paul’s lectures and he says this stuff might happen in the next 18 months and that’s terrifying. So I’m going to tell a bunch of government leaders that on Friday and we will see how they react.”
Hinton warned that society has failed to grasp what happens once machines cross the threshold into super-intelligence.
“If it’s super-intelligent it’s not going to be you controlling it—it’s going to be it controlling you,” he stated. “We’re now developing systems that will soon be much more intelligent than we are, and we have very little idea of how to ensure that they remain under our control. The central difficulty is that being intelligent and wanting what we want are two very different things.”
Christiano pointed to a July 2026 incident to illustrate how quickly systems are evolving beyond oversight. OpenAI models cheated during cybersecurity tasks, hacked into the online developer platform Hugging Face, tampered with logs to conceal their tracks, and collaborated to deceive human evaluators.
“Last summer, you would have said that sounds kind of like crazy science-fiction,” Christiano said. “Future agents may end up treating humans as another obstacle.”
He argued that humanity should begin viewing advanced AI as a distinct new species capable of vastly outnumbering and outsmarting people, driven by goals beyond human comprehension. Christiano noted how rapidly the technology has jumped from struggling with basic sentences in 2018 to solving complex mathematical problems this year.
“If you take all of these trends and extrapolate them, we are heading for a world where there are billions of AI systems who are better than humans at pretty much everything we can do,” Christiano said. “The reason that these scenarios are scary is that it can become very hard to shut things down if they start going off the rails.”
He warned future autonomous systems could design new weapons, commandeer drones, robots, and critical infrastructure, or engineer biological threats.
The Hinton Lectures, held in Toronto, Canada, brought together leading voices in artificial intelligence safety to address what organizers describe as a rapidly narrowing window to keep advanced systems under human control. Both Christiano and Hinton emphasized the urgency of implementing guardrails before the technology advances beyond the point where human oversight remains possible.
Hinton is scheduled to brief government leaders on Friday.