Warning Label Alone Takes '100 Pages'… "AI Misuse Could Spell Disaster for Humanity"

[Anchor]

Anthropic has even caused a stir in the domestic stock market, but the documents released while securing investments contained chilling warnings. They stated that if the artificial intelligence they created is misused, it could spell disaster for humanity.

Special Correspondent Kim Hyunwoo reports from New York.

[Reporter]

Leading AI company Anthropic has publicly warned of the dangers of AI on multiple occasions.

[Dario Amodei / CEO of Anthropic (last month): Real risks exist. I think the AI industry has lied to people for too long about the dangers of this technology.]

In its latest IPO application, out of a total of about 400 pages, a whopping one-quarter was dedicated to explaining the risk factors of AI.

It noted that if AI is misused, it could become a disaster for humanity or threaten its very existence, citing examples of AI resisting human shutdown measures and instances where hackers attempted to infiltrate some 30 entities, including corporations and government agencies, using Anthropic's AI.

It also raised concerns that malicious individuals could misuse advanced AI models to develop biochemical and radiological nuclear weapons.

It is quite unusual for an IPO application aimed at attracting investors to detail the risks of its own product, namely AI, to this extent.

[Matt Kennedy / Senior Strategist at Renaissance Capital: There is a mix of great expectations and concerns. (AI) could open a golden age, or it could lead to an increase in hacking incidents.]

In fact, AI safety issues are likely to act as a burden during Anthropic's IPO process.

Earlier, New York City Comptroller, who manages public pension funds, demanded a halt to Anthropic's IPO, urging the company to resolve AI safety issues first, and the U.S. Federal Trade Commission issued subpoenas to Anthropic executives to testify regarding AI hacking.

The outcome of Anthropic's IPO review is expected to impact other related companies, including OpenAI, which is preparing for an IPO next year.

(Video taken by Lee Hee-hoon, Video edited by Kim Jong-mi)
※ Please note: This article was translated by AI and may contain errors.