[DigitalToday reporter Yoonseo Lee] Dario Amodei (다리오 아모데이), chief executive officer (CEO) of Anthropic, warned that AI surpassing humans could give humanity transcendent power, but it remains unclear whether society is ready to handle it.

TechRadar, an IT media outlet, reported on June 3 that Amodei wrote in an essay earlier this year that the arrival of highly advanced AI could bring not only technological progress but also issues of limits in control and understanding.

The key is when AI capabilities surpass the human level. Scientists and the tech industry have consistently suggested AI could reach a stage where it creates new knowledge, acts on its own and improves itself. This assumption leads directly to discussion of existential risks. The concern is that if future AI becomes that powerful, humans may not be able to restrain it.

Amodei in particular said humanity as a whole may not even be ready to use, let alone manage, the technology it will create. "Humanity is on the verge of receiving unimaginable power, and it is still unclear whether we have the maturity to handle it," he said. The remark reflects a view that as the technology race accelerates, society's capacity to handle it could become a bigger problem than performance improvements themselves.

The direction he presented is closer to resetting the conditions for development rather than halting it. Amodei stressed the importance of pushing responsible AI development within a democratic system. At the same time, he argued that obstacles should be put in place to prevent authoritarian or hostile countries from developing their own form of AI. His message is that as the competition for AI leadership spreads into national-level strategies, not only the technology itself but also the developers and governance matter.

The remarks also intersect with changes in Anthropic's corporate standing. Anthropic has been rapidly growing its presence this year along with rival OpenAI. Separate from the company's growth, Amodei has long publicly discussed AI's direction and how society should respond, and recently his warnings have been intensifying.

A point that markets and the industry are watching is the timing of reaching AGI, artificial general intelligence. The race to develop AGI has already begun, and scientists believe AGI could emerge at some point as early as the second half of this year and over the coming decades. Sam Altman, CEO of OpenAI, has also repeatedly mentioned the advent of AGI.

Broader discussion is also becoming more specific. Computer scientist Ray Kurzweil forecast that AGI would be realised in 2029, followed by artificial superintelligence within the subsequent decades. The issue is not only when the technology is reached. If superhuman-level AI becomes a reality, whether society is prepared to handle the change remains a separate question.

The remarks are less about putting the brakes on optimism in the AI industry than about asking again what principles of control and political conditions should govern the growth of the technology as the development race intensifies. Amodei is stressing that more important than the power AI will create is who will handle that power and under what system.