Demis Hassabis, Chairman of DeepMind and Chief Scientist at Google, warned that failure by enterprises and governments to strengthen AI system security could escalate into a major security incident within the next few years. Speaking at an AI summit, Hassabis noted that while the industry has entered the era of autonomous agents with increased practical value, this shift has simultaneously elevated potential risks. He cited recent AI attack incidents as early warning signals, stating that significant consequences have been avoided so far only because current AI capabilities remain limited. Hassabis urged the AI industry to adopt safety frameworks from high-risk sectors like biology and nuclear technology to establish stricter safeguard mechanisms. Emphasizing that substantial work remains, he called for enhanced international cooperation and proposed the creation of an international standards body to formulate comprehensive safety standards for AI systems.