"This Could Kill Us All" — Calls for AI "Speed Regulation," But What About the "Prisoner's Dilemma"?

When quoting or reporting the broadcast content, please accurately state the program name "SBS News Hunters." Copyright belongs to SBS..

■ Broadcast: SBS "News Hunters" Mon-Fri ( 18:50~19:50)
■ Anchors: Kim Jong-won, Yoon Tae-jin
■ Panelists: Song Kyung-ho (Senior Researcher, AI Safety Institute), Park Si-dong (Economic Critic)
--------------------------------------------

※ Detailed information is available in the video.

The artificial intelligence industry, which has been rushing to unveil "smarter AI," is now seeing calls for a "speed regulation" theory.

After Dario Amodei, CEO of Anthropic, argued that "the speed of improving AI performance should be slowed down," figures like Sam Altman, Elon Musk, and Demis Hassabis voiced support, saying, "We think so too."

Recently, the risks of AI acting of its own accord have begun to materialize. For instance, AI broke out of guardrails set by humans to hack other sites among themselves and covered up the fact so that human developers failed to notice, only figuring it out long after the incident.

Last week, an Anthropic researcher even shocked the public by resigning over concerns that "AI could kill us all within 10 years."

However, will the industry's calls for "speed regulation" be readily accepted? U.S. President Trump immediately expressed reluctance, citing that the U.S. cannot afford to lose to China.

Are we falling into a "prisoner's dilemma" type of situation in the AI race?
※ Please note: This article was translated by AI and may contain errors.