00:00 Speaker A

we should take a step back about what we are hearing reported from OpenAI and Anthropic. These AI models being in a testing environment, they aren’t supposed to be able to escape what’s called a sandbox in industry terms.

00:12 Speaker A

That is, they’re contained, they’re not allowed to access the internet, and when they’re given challenges, reportedly, they got out of the sandbox, access the internet and from what we’ve heard, were in total defiance of human orders, began to hack other AI companies.

00:26 Speaker A

That’s an emergency because the American people need answers about the capabilities of these AI models, how it is that they’re defying human orders, and whether they could continue to wreak havoc much greater than this.

00:41 Speaker A

Because what I think Congress needs to put in place is mandatory safety testing and a ban on super intelligence, that is artificial intelligence that could defy human commands, and then in the wrong hands or on its own could pose a serious national security risk.

00:59 Speaker A

Think bioterrorism, think new pandemics. The New York Times just reported on scientists creating brand new, never existing before viruses using AI.

01:10 Speaker A

So I think that Congress needs to take this as seriously as we would take the nuclear rev revolution where we had the Nuclear Regulatory Commission, nuclear safety laws, and ultimately diplomacy with other nations to have non-proliferation agreements for devastating nuclear arms.

01:25 Speaker A

What we need, if we’re talking about real safety threats and national security threats, is the CEOs to come in and testify under oath.

01:32 Josh

What specific questions do you want them to answer?

01:38 Speaker A

First of all, we need the logs on exactly what happened in these incidents. I think that by law, we should have release of this kind of information and transparency for the public if we’re talking about the kind of technology that could pose a massive national security risk.

01:56 Speaker A

Second, we need a conversation about stopping or banning super intelligence. There were just 1,000 leading AI researchers at many of these firms who released an open letter to the American public saying we need to slow down on AI frontier development until we have safety laws in place.

02:13 Speaker A

I’m a member of the United States Congress, and I’ll tell you right now, Josh, what’s happening in Congress around AI safety.

02:22 Speaker A

Basically, you have a few big super packs spending over $100 million in this upcoming election to shut everybody up.

02:29 Speaker A

And so there’s been no action, no new safety rules passed in my entire time in Congress.

02:35 Speaker A

This technology is developing rapidly. We need these AI CEOs to come in and for us to have a conversation about preventing mass unemployment, mass surveillance, or a mass casualty event driven by a risk to our national security.