Advanced AI Could Trigger “Dangerous Consequences,” warns DeepMind CEO
Google DeepMind CEO Demis Hassabis says Artificial General Intelligence could be just five to ten years
prativad 
Google DeepMind CEO Demis Hassabis says Artificial General Intelligence could be just five to ten years away, but the road to building it won’t be risk-free. Speaking at the Axios AI+ Summit in San Francisco, he warned that advanced AI systems could lead to “dangerous consequences,” including cyberattacks on critical energy and water infrastructure.
Hassabis described AGI as a system capable of matching human cognitive abilities — from reasoning and long-term planning to creativity. According to him, today’s large language models are still “jagged intelligence,” smart in patches but lacking consistent reasoning and continuous learning.
He said a few major breakthroughs and more scaling could push AGI much closer. But the real danger, he stressed, may emerge even before AGI arrives. Cyberattacks on power grids and water systems, misuse of autonomous agents, and AI systems drifting from intended goals are all urgent risks.
His warning aligns with growing concern inside the tech world. A recent open letter signed by leading technologists — including Steve Wozniak, Geoffrey Hinton, Yoshua Bengio and Richard Branson — argued that superintelligent AI could threaten human freedom and even survival, urging a global pause on AI development until safety is guaranteed.
Not everyone sees doom ahead. Elon Musk recently said advances in AI and robotics could make work “optional” in the next 10–20 years, and even imagined a future where currency becomes irrelevant. But he also admitted that such a world still requires significant technological progress.
Where AI ultimately leads is still uncertain. What is clear is that the race toward AGI is accelerating — and the risks are no longer hypothetical.
