This isn’t an exaggeration.
Recommend this podcast episode with Ed Zitron and Brian Merchant:
The World of AI Regulation with Brian Merchant October 9, 2025 • BETTER OFFLINE Brian Merchant: There’s there’s one bill that Silicon Valley is genuinely upset about and afraid of. And it’s a bill that like sets the very lowest bar and it says, essentially, if you are going to make a chatbot and market it to children, then you have to be able to demonstrate that this chatbot isn’t going to make them harm themselves. Ed Zitron: Ah. So they hate that. Brian Merchant: They are and they’re in you know, they have there’s this Silicon Valley lobbying group that’s kind of famous in California especially, called the Chamber of Progress. It’s like, yeah, it’s. Ed Zitron: Fuck off. Sorry, just like immediate reflexive, Brian Merchant: I know, gag reflex immediately. So they they’ve got this guy who’s out there writing op eds in like the San Diego Tribune and doing press going Oh it’s over broad, And let me tell like their actual line on this is, if this bill passes, then you’re gonna be taking away AI that could educate children. You’re gonna be taking away AI from children, and they’re not gonna have the same advantages that children with AI have. And they’re running this like this big Facebook campaign. They’ve hired lobbyists specifically to… They often do this. Ed Zitron: It’s fucking evil people. Brian Merchant: They are evil people. I mean this, I mean, for me, this was like, you know, this was like it’s past the threshold once the Adam Raine stuff broke and and OpenAI is uh, you know, trying to hem in haw about you know, oh, well you know we’re going to do this or that it’s we passed it, we passed the rubicon, right There’s they’ve got chatbots that are that are telling children to you know, hide the noose so that their parents don’t see it. And like it’s again. They make it seem like, oh, AI is this frontier we’re gonna work something. It’s a it’s a product, it’s a software product. But yeah, go ahead. Ed Zitron: I have a theory, Yeah theory. Okay, So I don’t think they can control these models. I don’t mean because they’re intelligent. I don’t mean because they’re autonomous. I mean I don’t think you can actually prompt a large language model to categorically stop it doing something. Brian Merchant: Yeah. Ed Zitron: I don’t think it’s possible. Brian Merchant: Then you should not be selling that piece of software to children. Ed Zitron: Oh I fully agree. 100% agree. I’m just saying that I don’t think they’re capable. Brian Merchant: I think you’re right. Speaker 2 (14:30): But my theory is based on costs because of Claude code that they can’t that they can’t do cost control. If they can’t do cost control, it means the model won’t listen. And I reckon that they can’t be like never talk about killing yourself. Yeah, Like they just can’t do that. Brian Merchant: Yeah, or it would require like going back through you know, like there’s been I’ve seen a lot of sort of speculation that the reason that it’s talking like this is that it’s like a lot of the language is coming from like pro suicide forums in the bowels of the internet. So like they don’t want to go through and take the time to sort that out. Ed Zitron: Probably articles as well that say, how to deal with someone who’s – just horrible stuff. These fucking people. And is this a California bill? Brian Merchant: Yeah, it’s a California bill. Yeah, and it’s and it is Yeah. So you know, if you’re on Facebook there so that there’s a there’s the front group spun up by some of the VC firms like a16z and you know, Andreessen and these guys and YCombinator. Then there’s a front group called the American Innovators Network, and they’re running all these ads arguing that this bill, again whose sole purpose is to ban chatbots from being marketed to children, that that also try to convince them to harm themselves.
On this podcast they also talk about how it was the country music industry that thwarted the federal legislation to ban state level AI regulation.