Google DeepMind CEO Demis Hassabis has issued a stark warning to world leaders: the window to prepare for Artificial General Intelligence (AGI) is closing fast, and robust safety rules must be established before more powerful systems are deployed.
In a recent essay, Hassabis emphasized that AGI—an AI capable of performing any intellectual task that a human can—may become a reality within just a few years. He called for an international body to test advanced AI models before release, arguing that proactive regulation is essential to mitigate risks.
What Is AGI and Why Does Hassabis Want Caution?
Unlike today’s narrow AI, which excels at specific tasks like writing or coding, AGI would handle diverse challenges, learn autonomously, and solve problems much like a human. Hassabis wrote that AGI is “just a few short years away” and urged governments, researchers, and companies to collaborate rather than compete without safeguards.
“We’ve already seen the challenges frontier models pose for cybersecurity, and other threats, including nuclear and bio risks, may soon emerge as capabilities continue to advance,” he said.
Why AI Safety Matters More Than the Race for Better Models
The AI industry has prioritized speed and capability, but Hassabis argues that safety deserves equal attention. As AI grows smarter, the potential damage from mistakes or misuse escalates. He advocates for global safety standards, independent testing, and regular reviews to ensure responsible development without stifling progress.
The World Needs to Prepare Together
Hassabis stressed that countries must act now—not later. Governments need updated laws, businesses must exercise caution, educational institutions should teach AI-relevant skills, and cybersecurity teams must anticipate new threats. “The world still has a chance to get ready. However, it may not last for long,” he concluded.