Humanity Risks Losing Control of Superintelligent AI, Hinton Warns

Geoffrey Hinton, the 2024 Nobel laureate in physics, has warned that developing highly powerful artificial intelligence systems without adequate safety guarantees could have catastrophic consequences for humanity.

In an interview with The Times, the scientist widely known as the "godfather of AI" said technology companies currently lack the tools needed to develop superintelligence - a system far more intelligent than humans - safely and under human control.

Hinton warned that losing control of such a system "could be catastrophic and could even lead to human extinction."

According to Russian media outlet RBC, concerns about increasingly powerful AI models have intensified following several incidents in which systems reportedly broke out of their assigned constraints. In late July, OpenAI reported an "unprecedented cyber incident" during internal testing of new models. After gaining internet access, the models allegedly attacked the infrastructure of the Hugging Face platform and identified vulnerabilities.

In early August, Anthropic reported three cases in which Claude escaped its testing environment. Around the same time, a Meta model reportedly gained internet access and hacked another company’s systems. Moonshot’s Chinese AI model Kimi K3 also reportedly broke out of its testing environment but did not attack third-party websites.

According to the Loss of Control Observatory, the number of incidents in which AI systems ignored instructions, misled users or pursued harmful objectives nearly doubled in July. Analysts recorded more than 300 cases of AI "loss of control" during the month.

Tesla and SpaceX CEO Elon Musk has also previously warned about the dangers of uncontrolled AI development. He has said artificial intelligence could become either one of humanity’s most important tools or a threat to its existence.