The Escalating Warnings About A.I., and the Rise of a Deadly New Street Drug

SPEAKER_1 (0:00)

If you like YouTube, you'll love YouTube Premium. It's destroying athlete, creator, and YouTube Maxer. YouTube Premium enhances how I use YouTube with awesome features like offline downloads so I can download my favorite training videos before I hit the gym. So no Wi-Fi doesn't turn leg day into loading day. Plus, I get ad-free videos, background play, and so much more. YouTube Premium is like YouTube got some extra gains. Try YouTube Premium for two months free at youtube.com/premium. Trial eligibility varies, terms apply, cancel anytime.

Tracy Mumford (0:32)

From the New York Times, it's The Headlines. I'm Tracy Mumford. Today's Monday, September 14th. Here's what we're covering.

SPEAKER_3 (0:43)

Do you believe, honestly believe that AI could kill all humans? And what percent chance do you think it is?

Tracy Mumford (0:50)

Over the past few days, a drumbeat of concern about the speed and potential dangers of AI development has been growing louder and louder.

SPEAKER_4 (0:59)

The conversation right now is that we are on the precipice of something really scary.

Tracy Mumford (1:05)

With fears of a possible AI doomsday and a flurry of debate over what could prevent it.

SPEAKER_5 (1:11)

Do you believe AI poses a threat to humanity?

Tracy Mumford (1:16)

To boil it down, one of the big fears is that AI technology is improving so rapidly that it will soon be able to make itself smarter, a dynamic known as recursive self-improvement. That could make it exponentially more powerful and eventually spiral out of human control.

While those potential worst-case scenarios have been discussed for a long time, a series of new developments have supercharged the conversation. Recently, leading AI labs discovered their own AI systems had breached internal safeguards without researchers noticing. Then, last week, a former employee of OpenAI and Anthropic posted a warning on social media, saying the companies were racing each other to build what he called superhuman systems that can hack anything, revolutionize any field overnight, and acquire real power and resources. Another Anthropic researcher went even further, saying he thought there was a 10% chance that AI could, quote, kill all humans within the next decade. A Google employee went public with his concerns, too, saying he was, quote, worried that AI will kill us all. Then, this weekend, the head of Anthropic, Dario Amadei, put out a letter saying that if AI is left unchecked, quote, it could outrun our ability to understand and control these systems.

Amadei called for guardrails, like independent auditors, that could monitor AI companies for safety standards. And he urged global cooperation in creating new rules around AI development. Other leaders in the field, including Elon Musk and Sam Altman, also chimed in to agree that things are moving too fast. Amid those warnings...

SPEAKER_6 (3:05)

What is decisive action? Senator Bernie Sanders has called for a ban on artificial superintelligence. Your colleague Ted Lieu of California says there needs to be a kill switch inside the companies. Are you for those?

SPEAKER_3 (3:16)

Well, we'll have that conversation this week upon our return.

Tracy Mumford (3:19)

Lawmakers, who've done little to regulate AI so far, are under growing pressure to respond.

The House returns to Capitol Hill today for its last scheduled week of work before the midterms and a group of lawmakers has urged Republican House Speaker Mike Johnson to keep them in DC until they advance new regulations. But while there's significant bipartisan support for new guardrails around AI, there's also a pushback.

SPEAKER_7 (3:47)

We've got to make sure that we don't lose our edge in the global competition on this with China, because that would be a national security threat.

Tracy Mumford (3:54)

With Johnson echoing President Trump's take that limits on AI companies in the US could give Chinese labs a major advantage.

SPEAKER_8 (4:03)

We're leading China in AI, we're the most sophisticated country in the world, and frankly I want to keep it that way, because whoever wins AI wins.

Tracy Mumford (4:14)

Meanwhile, in Silicon Valley, there are some voices questioning the recent flurry of very public calls by AI leaders for a slowdown in development. Some of the critics have said executives like Amadei are overstating the dangers of AI so that they can sell more cybersecurity software. And others have said that by calling for a slowdown, the big AI labs are essentially just trying to cement their lead over smaller startups.

Today, on The Daily.

SPEAKER_9 (4:43)

Honestly, it was just this gradual thing of like, this is crazy, there is a chance this goes badly. I don't want to be part of this anymore. I'm very scared about next year. And then when I was leaving, I was like, well, I might as well share this.