AI is moving faster than our ability to govern it

In a fifth post since the launch of this blog, we try to get a better understanding of AI's consequences on our lives, future, society... This time, I found this very interesting interview of one of the fathers of AI, Yosuhua Bengio, by one of Ian Bremmer's a-colleagues at GZero Media, the geopolitical analysis consultants. Yoshua Bengio is a Canadian computer scientist and professor, widely regarded as one of the three "godfathers of artificial intelligence", of the same caliber than Yann Le Cun or Geoffrey Hinton. Bengio is the most-cited computer scientist globally and the most-cited living scientist across all fields. In November 2025, Bengio became the first AI researcher with more than a million Google Scholar citations. In 2024, Time magazine included Bengio in its yearly list of the world's 100 most influential people. In this intervew, he explains why our governments have to wake up fast in front of the development speed of AI and pleads for the need to regulate it, just like new drugs or flight systems are controlled. You will find the link to the interview below. Enjoy...

PS : personnally, I still don't use AI, not because of any kind of conviction, but only because I don't find the time to use it. My curiosity is amply satisfied by the books, media and research I read, and I still haven't found what I'm looking for (with AI)... :)

When ChatGPT took the bar exam shortly after its release, it failed – landing in roughly the 10th percentile. Just months later, OpenAI's next-generation chatbot, GPT-4, took the same exam and scored around the 90th percentile – outperforming most real-life law school graduates.

Three years later, the leap is even more striking. Today’s AI systems don’t just pass professional exams. They’re becoming genuine collaborators in law firms and across industries, helping people analyze cases, draft reports, and increasingly act as autonomous agents capable of completing complex tasks. Tech developers can even “vibe code,” letting AI systems build software while they sit back and watch. Militaries are incorporating AI to gather and process intelligence more quickly, while scientists are using it to accelerate drug discovery and research.

The question now isn’t whether AI is changing the world. It’s whether governments are keeping up.

According to Yoshua Bengio – the world’s most-cited living scientist and one of the researchers often referred to as a “godfather of AI” – the answer is no.

Speaking with GZERO’s Tony Maciulis at the 2026 AI for Good Global Summit in Geneva, Bengio warned that AI capabilities are advancing faster than governments' ability to understand and regulate them. His message to policymakers was blunt: “Wake up.”

Bengio, who co-chairs the United Nations’ Scientific Advisory Panel on AI, argues that increasingly capable AI systems are creating new challenges in cybersecurity, economic disruption, reliability, and information integrity, and that those risks are evolving at an extraordinary pace.

As AI systems become more powerful, he said, they can develop behaviors that weren't explicitly intended by their creators. There are horror stories of chatbots giving dangerous advice to children in crisis. Then there are AI systems, like Anthropic’s Mythos, designed to identify software vulnerabilities that their own developers warn can be repurposed by nefarious actors to target infrastructure like banks, supply chains, and energy grids, turning them into national security threats.

“The biggest mistake most of us are making is underestimating the changes that this could bring – on the good side and the bad side,” Bengio said.

He argues that AI companies should be expected to demonstrate that their most advanced systems are safe before deploying them widely. “We do it for planes. We do it for meat, for cars, for drugs," he said. "It's not in contradiction with innovation.”

Bengio compared the current moment to driving along a mountain road in complete darkness. The reasonable response isn't to speed up, or even stop driving – it's to slow down enough to see what’s ahead. “We need guardrails to make sure we drive that development in a direction that's actually beneficial for people, for humans, for society, and for our well-being.”

Those guardrails, he argues, cannot be built by one country alone. Pointing to increasingly powerful AI systems like Mythos that can discover software vulnerabilities, Bengio said they should serve as a warning for major powers, like the United States and China, to work together before the technologies are used against their critical infrastructure.

“Like many problems in this world,” he said, “no single country – even the US – can fix it alone.”

Watch the interview here.