ML Matt Levine Writes Money Stuff, the daily Bloomberg Opinion column on Wall Street, securities law and the stranger corners of financial markets: meme stocks, crypto structu...

“I think that it is not a risk that is really cognizable in the risk factors of an IPO prospectus because if they kill everyone on earth, they won't get sued for securities fraud. So, who cares?”

Money Stuff: The Podcast · October 2026

“I think that it is not a risk that is really cognizable in the risk factors of an IPO prospectus because if they kill everyone on earth, they won't get sued for securities fraud. So, who cares?” — Matt Levine, Money Stuff: The Podcast

Context

The risk in question is an AI company's own product killing everyone. Levine and Mary Childs were talking about whether Anthropic's IPO prospectus has to say so. Levine thinks it does, because worrying about that is the company's whole identity and leaving it out would look odd, but he doubts it works as a risk factor in the legal sense.

Transcript

Money Stuff: The Podcast Around 37:25 into the episode
Matt Levine

…although their models have a control. You know what I'm asking? Because at various times, like he was, you know, founder of OpenAI. Like he's at various times been very worried about the existential risk of AI. At other times, been pretty blasé about it. Don't really know what's going on there.

Mary Childs

Well, it's before the playa and after, obviously. I'm just kidding.

Matt Levine

Are you? But SpaceX did not say in its prospectus, we might kill everyone on Earth, even though it is in theory an AI lab, as well as a space rocket lab. And I think people were not that. Surprised by that because, like, the culture of SpaceX is not really built around worrying about AI killing everyone. Whereas Anthropic is their whole thing. And so, if they left it out of the prospectus, it would be weird. So, they have to say it. But I think that it is not a risk that is really cognizable in the risk factors of an IPO prospectus because if they kill everyone on earth, they won't get sued for securities fraud. So, who cares?

Mary Childs

But I liked your footnote that was maybe other governments will clamp down if they think that we really are about to destroy everyone. And that is a business risk.

Matt Levine

Yeah, I am kind of kidding, right? Like, the point is that if you really think that AI could kill everyone on earth, like one possibility is there'll be like a step change like next week and we'll all be dead and then oops, right? But like, the other possibility is like there's a lot of bad things…

Mary Childs

of hits to profit, I'm hearing. Yeah. Hacking

Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.

More from Money Stuff: The Podcast