Sam Altman: "A monopoly on AI power would be a catastrophe for humanity… We need an AI regulatory body"

Sam Altman, CEO of OpenAI, on the 18th (local time) publicly stressed the need to build artificial intelligence (AI) governance at the "4th AI Impact Summit" held in New Delhi, India. He also said an "international AI regulatory body" similar to the International Atomic Energy Agency (IAEA) should be established.

Speaking that day, Altman said, "The world urgently needs to build a dedicated regulatory body in the AI field like the IAEA, which monitors nuclear proliferation," adding that "with today’s institutions and pace, we can’t keep up with technological progress." He argued that if powerful AI technology and resources become concentrated in a handful of companies and a handful of countries, "it could create a risk that could drive humanity to catastrophe," and he pointed to the need for transnational regulation and a joint monitoring framework to mitigate that risk.

He also projected that AI could advance faster than expected. Altman said, "Within a few years, we could reach an early form of true superintelligence," and predicted that "around 2028, a significant portion of the world’s intellectual capacity may exist not in human brains but inside data centers." He said so-called superintelligence would outperform human scientists and executives across a range of areas including research, management and strategy-setting, and emphasized that the international community must first design safety guardrails and accountability structures before AI at that level emerges.

With international discussions over AI regulation and governance gaining momentum, Altman’s simultaneous public mention of an "IAEA-style AI oversight body" and the possibility that superintelligence could arrive within years is seen as a message to both markets and policymakers.

Altman highlighted the political and economic risks that could come from a concentration of AI power, putting the concept of "democratic AI" front and center. He defined "the democratization of AI" as "the best path for humanity to prosper." He warned that "AI becoming overly centralized in one company or one country can pose a major risk to all of humanity." Emphasizing that AI should be designed to expand freedom and human choice, he presented three pillars to achieve that: AI access, real-world use and adoption, and user agency. "AI should be a productivity tool for the many, not a means of control for the few," he said, repeatedly underscoring the dispersion and openness of technology and interoperability.

Asked about the labor market and the real economy, he maintained cautious optimism. Altman acknowledged that "it’s an undeniable fact that AI will have a clear impact on existing jobs," but said that, as has historically been the case with automation and technological innovation, "I remain confident that new and better jobs will be created." However, he warned that if safety and trust are not secured, AI could be seen not as an engine of economic growth but as a source of social instability.

By Eun-i Ko koko@hankyung.com