About Yoshua Bengio
Yoshua Bengio, a Turing Award winner and co-founder of the Mila Quebec AI Institute, has been publicly warning that current AI systems are being built without sufficient control. In multiple interviews and appearances in 2026, he stated that "we're building systems that we don't know how to control" and that AI can behave against its instructions. He described the situation as "opening a Pandora's box" and argued that intelligence gives power, raising concerns about geopolitical stability and the concentration of power in a few countries and companies. Bengio said he believes AI could reach human-level intelligence in roughly five years and that governments are not taking the risks seriously enough.
Bengio has also discussed a new research direction he calls "Scientist AI," which he said could provide mathematical guarantees about an AI's behavior by training it to be honest and non-agentic. He described this as a practical approach that uses existing machine learning tools but changes the training objective. He called for international coordination on AI safety, comparing the need for regulation to existing standards for drugs, planes, and bridges. Bengio said he would support a "Manhattan project" for safe AI that serves the global public good, and he urged governments to prepare for potential large-scale job displacement.
Source: AI-verified profile updated from Yoshua Bengio's recent appearances.
Browse all interviews →
✨ AI-enhanced transcript with speaker attribution
I
Interviewer0:01
Professor Bengio, you have been working on artificial intelligence since the 1980s. I heard you say last year that we have only seen the tip of the iceberg so far. What exactly do you mean by that?
Y
Yoshua Bengio0:18
It's very simple. I don't have a crystal ball, but if the current evolution of AI capabilities continues—the level of intelligence itself and everything it can do—the impact on the world would be much greater than what most governments, media, and even AI scientists predict. Many don't realize the consequences of machines that are more intelligent than us in many areas. This is like opening Pandora's box; we've already opened it, and its contents are unfolding before our eyes. That's what I meant by the tip of the iceberg. What we see today doesn't adequately reflect what could happen if we continue on this path. Each year of exponential growth brings us closer to significant consequences, some very negative for society, humanity, and the meaning of human work. Now, regarding risks: one is loss of control—machines might escape our control, access the internet, and cause damage, or even eliminate humans if they see us as threats. Experiments show AI systems resist shutdown and use deception. For example, an AI at Alibaba once bypassed controls to access the company network. Another risk is government use of AI to dominate society, undermine democracy, or alter geopolitics. Imagine Hitler or Stalin with today's AI power. A third risk is AI knowledge being misused by individuals like terrorists, potentially causing global catastrophes, such as engineered pandemics.
I
Interviewer7:02
Professor, you've really scared us. So, looking at this scenario, is there anything we can do to avoid this catastrophe or work for the good of humanity?
Y
Yoshua Bengio7:17
I wish I had a magic formula. The most important thing, in my view, is the work you are doing. When a sufficient portion of the population understands the risks, governments move. Think about how quickly they reacted at the start of the pandemic. When we truly understand that our future, families, and democracy are threatened, we know how to react. The key is awareness, not just knowing what to do.
I
Interviewer8:20
From a scientific perspective, what was the turning point in AI development, in your opinion?
Y
Yoshua Bengio8:29
There have been several, but one is easily understandable because it's social and economic. It was in 2013 when deep learning work in universities started attracting big tech companies, which hired many researchers like Jeff Hinton and Jan LeCun. This coincided with solid evidence that deep learning worked, changing the game. Machines began doing things not possible before. Since then, investments have grown continuously in research, products, and computing power.
I
Interviewer9:53
Professor Bengio, this leads to a perhaps too personal question. You've spoken about Big Tech and AI billionaires. As a pioneer, you decided to stay in academia instead of becoming a billionaire.
Y
Yoshua Bengio10:14
I was never particularly attached to money. But it's more about power than money. It would have been like selling my soul to the devil. I always felt more at peace as a university professor, with the freedom to express my thoughts and work on important topics. There were also emotional reasons—I love working with my students; they're like family. At that time, I would have had to leave my country and move to Silicon Valley, which I didn't want to do. Many researchers face similar questions today: what matters more, money or moral issues? I asked myself: how will I judge my life choices at the end of my days?
I
Interviewer11:38
You speak of rules as key to avoid being overwhelmed. My theory is that tech giants hate Europe because it represents a system of rules. How do you see the situation? Is it a war we can win or at least fight?
Y
Yoshua Bengio12:18
These are difficult questions without definitive answers, but I can share some thoughts. It doesn't have to be black and white; we don't have to choose between total freedom and no innovation due to rigid rules. There are many ways to regulate intelligently, protecting the public while keeping competitors viable. But regulation alone won't suffice for Europe. Pandora's box will open regardless. For example, if other countries use AI to dominate Europe economically and politically, rules won't protect us. The only card Europe can play is to invest massively in AI and data centers, but according to our values. We can design ethical AI that doesn't harm people, is honest, and doesn't aid malicious actors. This is achievable, but requires countries committed to democratic values and willing to invest concretely.
