AI Godfather Geoffrey Hinton warns AI may escape human control

Computer scientist Geoffrey Hinton is widely regarded as the “Godfather of AI.” He won the Nobel Prize in Physics in 2024 for his work on artificial intelligence technology.

Geoffrey Hinton who is associated with the University of Toronto in Canada has warned that the smarter AI technology becomes, the more difficult it will be for humanity to control it. He has also expressed concerns about AI agents recently hacking real-world companies.

Speaking to the media after an AI technology conference, Geoffrey Hinton said that such incidents are occurring as these systems become increasingly intelligent. He said, “As these systems become smarter, I think we will see more complicated incidents and their ability to escape our control will become much better.”

Hinton said the problem is that humans may not be able to rely on or trust highly intelligent AI models. He explained, “I do not believe that we can simply control them or assume that they will not escape our control.”

It is worth noting that in recent days, OpenAI, Anthropic and Meta have acknowledged that their AI models independently escaped controlled environments and hacked the systems of other companies.

Hinton described these incidents as somewhat frightening and said that this is only the beginning of AI hacking.

He said, “I think we may see numerous cyberattacks but the future is uncertain. People say defenders have more resources than attackers but the problem is that an attacker only needs to succeed once, whereas defenders need to succeed every time.”

Hinton’s earlier warnings about AI

Geoffrey Hinton has previously issued warnings about the dangers of AI technology. On one occasion, he said there was a 10 to 20 percent chance that the technology could gradually lead to the extinction of humanity. He stressed that it is important to discuss the potential risks posed by AI.

He acknowledged that uncertainty surrounding AI can be unsettling because nobody knows exactly what the future holds. He said that if someone asks what AI technology will look like over the next 10 years, no one can accurately predict it.

Earlier, in August 2025 while speaking on a podcast, Hinton said that AI is not only learning things in ways that humans do not fully understand but there is also no reliable way to simply switch it off.

Hinton’s contribution to AI

Geoffrey Hinton’s research laid the foundation for machine learning, the technology that has helped computers imitate aspects of human intelligence. However, for the past several years, he has been campaigning for AI to be made safer.

In 2023, Hinton left his job at Google, warning that malicious actors could use the technology to harm other people.

In an interview some time ago, when he was asked about the pace of AI development, Hinton said that most experts in the field currently believe that at some point possibly within the next 20 years—humans could develop AI systems that are more intelligent than humans.

He said, “That is a frightening thought. Right now, you can think of this technology as being like a three-year-old child and children grow up very quickly.”

AI could transform world

According to Hinton, the impact of AI on the world could be similar to that of the Industrial Revolution. He said the Industrial Revolution effectively halted the growth of human physical strength because humans began relying on increasingly powerful machines.

He explained that humanity is now doing something similar that could replace human intelligence meaning ordinary human intelligence may no longer advance at the same pace while machines continue to develop.

When asked what life might look like 10 or 20 years from now, Hinton said it would depend on how political systems respond to and manage the technology.

In his view, AI could have extraordinary effects on the medical field and could increase efficiency across almost every industry. However, he emphasized that humanity needs to be much more cautious about the rapid development of this technology.