- “ Ai godfather ‘yoshua bengio warns that the breed of AI prioritizes speed on safety
- This risks unpredictable and dangerous consequences
- He urges global cooperation to apply AI regulations before autonomous systems become difficult to control
“ Ai Godfather ” Yoshua Bengio has helped create the foundations of neural networks that execute all kinds of AI tools today, chatbots imitating cartoons to scientific research assistants. Now he has an urgent warning for AI developers, as he explained in an interview with Sky News. The race to develop increasingly powerful AI systems is at a rate which, in its opinion, is far too reckless.
And it is not only what company builds the best chatbot or who gets the most funding. Bengio believes that the rapid and unregulated push towards an advanced AI could have catastrophic consequences if security is not treated as an absolute priority.
Bengio has described watching the developers run against each other, becoming botched or taking dangerous shortcuts. Although the speed can make the difference in the break of a new type of product worth billions and to make up for a rival, it may not be worth it for society.
This pressure only intensified for AI developers with the rise of Chinese AI companies like Deepseek, whose advanced chatbot capacities have drawn the attention of Western and governments. Instead of slowing down and carefully considering risks, large technological companies accelerate their development of AI in a total sprint for superiority. Bengio is worried, this will lead to precipitated deployments, inadequate security measures and systems that behave in a way that we do not yet understand completely.
Bengio explained that he had warned the need for stronger AI surveillance, but recent events have made his message even more urgent. The current moment is a “turning point”, where we implement significant regulations and security protocols or may let the development of AI relax in something unpredictable.
After all, more and more AI systems are not only processing information but can make autonomous decisions. These AI agents are able to act alone rather than simply respond to user entrances. This is exactly what Bengio considers the most dangerous path to follow. With enough computing power, an AI that can develop strategies, adapt and take independent measures could quickly become difficult to control if humans wanted to take over the reins.
Takeover IA
The problem is not only theoretical. Already, the models of AI make financial transactions, managing logistics and even write and deploy software with a minimum of human surveillance. Bengio warns that we are just a few steps from the much more complex, potentially unpredictable AI behavior. If a system like this is deployed without strict guarantees, the consequences could range from boring hiccups to full and economic crises.
Bengio does not call the development of AI. He clearly indicated that he was an optimist on AI’s capacities when used in a responsible manner for things like medical and environmental research. He simply sees the need for a change in priority to more thoughtful and more deliberate work on AI technology. His unique perspective can gain some weight when he calls AI developers to put ethics and security before competition with rival companies. This is why he participated in political discussions during events such as the next International Summit of IA Security in Paris,
He also believes that regulations must be reinforced by companies wishing to take responsibility for their systems. They must invest as much in safety research as in performance improvements, he says, although this balance is difficult to imagine appearing in the AI mele of today. In an industry where speed is equivalent to domination, no business wants to be the first to hit the brakes.
World Cooperation Bengio Pitchs may not appear immediately, but as the AI’s arms race continues, the warnings of Bengio and others in similar prestige positions become more urgent. He hopes that industry will recognize the risks now rather than when a crisis will force the question. The question is whether the world is ready to listen to before it is too late.
