California Regulates AI Chatbots Amidst OpenAI's Bold Moves

Welcome to the AI Applied Podcast. Today on the show, Connor and I are talking about some big changes, some legislation that is coming to AI. We have two different kind of pieces of news that we think are pretty relevant. One is that Sam Altman recently announced that if you pass age verification for adults on ChatGPT, then you can start having erotica content on the app. And this is something that is like pretty 180 from all of OpenAI's policies before this. And at the same time, you have California, which has just become the first state to regulate AI companionship chatbots. And there's a bunch of interesting things in there. So today on the show, we're going to kind of talk about the state of AI safety and AI regulation today. Connor, right off the bat, is there any parts of the story that you're dying to get to? I know you've been posting about this on LinkedIn, has been blowing up. There is a furious debate going on. So I'm curious your take overall on this. What a weird week. You know, it was just such a whiplash kind of week, right? So it was on Monday, October, I'm not even probably maybe 13th or something like that. When Gavin Newsom, governor of California, announces that SB 243 is going into effect or will go into effect January 1. And this is just, I thought, I don't know. I don't want to say sort of like universally great, but I can't see any problems with it. Jaden, you're a dad, I'm a dad. But this is one of these things where I think parents were probably pretty happy. The idea is that it requires chatbot operators, right? So, you know, OpenAI or anybody, whatever it is, to essentially implement safeguards around, you know, how people interact with these things, right? So, like, they have to prevent chatbots from exposing minors to sexual content. We remember, wasn't it Meta a while back that was saying this is okay for over 8 years old and over 12 years old? And we were just, like, grossed out by it. Anyway, so it requires, like, notifications, reminders. Reminders, by the way, who cares about reminders? But still, like for miners that chatbots are AI generated. Although in that case, you know, Jaden, it's funny, like usually reminders like, oh, whatever. If every once in a while a chatbot has to say, by the way, I'm just a chatbot, I'm not alive. I think that goes a very, very, very long way. It may be overlooked, like reminders to sort of say, be careful. But a reminder from the chatbot itself is huge. The law itself is all about, you know, suicide prevention protocols, identity disclosure, meaning like they have to issue a notification at the beginning of any interaction saying, hey, remember, I'm AI. But again, I'm all for this thing, continuously reminding minors of that. There's also reporting requirements. There's legal accountability, which provides families with the right to pursue legal actions, et cetera. The thing that, okay, so that's on day one, right? And by the way, the funny thing about this is that I was like getting all like, this is awesome. Open AI has been great around this. Open AI, Sam Altman, we've heard him speak a ton about, hey, we really want to prevent people from harming themselves. Finn, who's 16, who I work with a ton, that's my son, he and I have tested things out where he's like, hey, I'm 16 and this is, you know, we sort of like do it in temporary memory so it doesn't raise any red flags, but sort of like starting to give indications of, oh, I'm feeling like this. And Open AI was doing a great job in the chat of saying, hey, listen, if you're having problems, blah, blah, blah, let's just stop this right now. I was like very impressed with it. but then jayden the next day has that legendary tweet where he's like hey guys blah blah blah stuff stuff stuff and also just because you know adults are adults we're gonna start letting chadjapiti do erotica anyway see you tomorrow everybody like so jayden you see this and you thinking what because on one hand open ai is like yeah we want to be careful We won be super guardrailed because they are in enterprise OK, like and this is where the big money comes from, from enterprise. And then enterprise has to see and wake up the next day to also erotica. Thoughts around this? Yeah. So, I mean, at the end of the day, I think it's just like a decision about where your brand wants to go. A lot of people in the argument are like, well, you know, it's a public company can do whatever it wants. I mean, yes, public companies can. It's kind of impressive. It's at, you know, 800 million weekly active users. I think that there will be like repercussions. So basically, like the news that I'm seeing from a lot of different places. So there's this TechCrunch article, for example, it says, ChatGPT's mobile app has seen slowing downloads growth and daily use analysis shows. Okay, so OpenAI. And then again, there was also another article that came out that was saying OpenAI has made about $1 trillion worth of commitments. So to me, I think it's honestly, this is all just about money. Basically, OpenAI has to fund a trillion dollars worth of commitments to build out their compute deals with Oracle, with, you know, NVIDIA. They have some deals with AMD. They have a bunch of deals. They're making a Broadcom. They just signed a deal. Right. So they have these multi, you know, hundred billion dollar deals. They've signed with everyone a trillion dollars total. And they have five to ten years for all of these deals to go through. So they have to spend a trillion dollars the next five to ten years. So that's pretty insane. and at the same time so much money so at the same time though we're seeing you know things like okay their mobile app downloads is slowing and there's you know more competitors coming from google and anthropic and at the same time um like they want to keep it up so i think we it feels like open eyes doing all of the plays everything that they're getting criticized for uh i don't think is like we've just decided that the world is ready for xyz so we're gonna do it just on a you know just purely principled standpoint. And this is kind of the way Sam Altman comes across. I don't buy that one spec. Their app downloads are stalling. They have to spend a trillion dollars. They want as much users as possible. And I think Sam Altman, literally one month ago, or maybe six weeks ago, there's a clip of him making fun of a grok in an interview when they were like, you know, is there anything that like you have thought about doing that you haven't? And he's like, well, you know, sometimes we get tempted to like sidetrack from our mission and do things that are off you know off of our mission because we think we'll get you a lot of users like oh like what and he's like oh well you know like we haven't added a sex bot avatar to chat you're like oh it's like kind of this awkward moment in the interview or whatever i get that but at the same time that was like six weeks ago and now he's like okay actually just kidding we changed our mind and we are doing it so to me i think this is just purely financially motivated he saw that grok elon musk's xai got a lot of like news and hype around all of his like edgy chat bot avatar things oh my gosh and so it feels like he's just trying to jump on the same bandwagon to try to spur users i would even go so far as to say that like sora 2 and some of the deepfake cloning stuff that they did there that has gotten a lot of um like pushback and negative press also was just i'm trying to get more users they're just trying to like push out anything that can go as viral as possible and get more users as possible because they just need more money so that's that's my opinion on basically where all this is coming from. Yeah, no, I mean, it makes sense. I guess I'm just thinking the struggle in my mind is what is this thing? And I'm not to get sort of like, we should start off like AI applied, like go back to like episode one, like what is AI and what's it going to be and everything else? Because I think different people think of it as very different things. So if you're an enterprise, you know or if you looking to I don want to say make money off it but if you looking to integrate it into enterprise and I work with a lot of companies obviously that have spent a lot of money on ChatGPT Enterprise you don want any of this stuff entering any of this at all right I mean you don't want anything like this inside your company. You're thinking, what is OpenAI doing? My gosh. I mean, the whole point is like, we just paid them $8 million, and we're just one of however many companies, and all we want is for this to be more productive. You're talking about how it's going to revolutionize the world and revolutionize work and all that kind of stuff. And then out of nowhere, it's like, by the way, if you're over 18, which people inside Enterprise are, you can do erotica too. And you're like, wait, what are you talking about? Because, you know, even on the Internet, if you are on, you know, there's like firewalls and things like that. Like you can't just like turn on pornography and things like that, obviously, in your enterprise space or anything like that. But my question is then for everybody else using it, because most people who are using this, it's something like 75% are using it for personal use. We've seen that, right? And, you know, and you and I talked to Mustafa Suleiman over at Microsoft. You came to New York. We had that amazing conversation with Mustafa. And he talked very little about work at all, right? I mean, he was talking about this as a whole new way of living. As, you know, Copilot, which is a phenomenal product, I think, using Copilot as a new kind of companion for life. Part of that is work. But, yes, but a lot of that is just how do you just live your life? This goes well beyond what work is and much deeper into how do people just live in anything, right? So then Sam comes back with this. Can I just like read this tweet real fast? Because this is amazing. He says, okay, this tweet about upcoming changes to ChatGPT blew up on the erotica point much more than I thought it was going to, exclamation point. It was meant to be just one example of us allowing more user freedom for adults. Here's an effort to better communicate it. And so it's worth reading this. Okay. As we've said earlier, we are making a decision to prioritize safety over privacy and freedom for teenagers. And we are not listening to any policies related to mental health. Okay. I think that's important. This is a new and powerful technology. We believe minors need significant protection. Yeah, man, because like this is like what separates us from the animals. And like when you see like what Facebook did years ago, nobody wants to repeat that. Right. So anyway, I continue. But we also very much care about the principle of treating adult users like adults. as AI becomes more important in people's lives, allowing a lot of freedom for people to use AI in the ways they want to as an important part of our mission. It does not apply across the board, of course. For example, we will still not allow things that cause harm. We won't treat users who have a mental health crisis different, blah, blah, blah, without being paternalistic. We'll help people, blah, blah. But we are not the elected moral police of the world. In the same way that society differentiates other appropriate boundaries, our rated movies, for example, we want to do a similar thing here. Jaden, here's the problem. The problem is what is ChatGPT for? because I'm telling you, sorry to be a broken record here, but there's people out there, enterprises out there who Sam Altman is trying to convince massive enterprises. Barclays just bought, what was it? Like a hundred thousand licenses of co-pilot. There's enterprises out there that are thinking AI is a replacement for how people work. And that's every, that's a $18 trillion business or market. And then there's other people who are saying like, no, this is just something I talked to sort of like as a companion, which one is it? Cause it can't be both. And I'm not saying he's not trying to thread the needle here. But I am saying that if you just like sort of willy-nilly go out and say there's erotica here, and all of a sudden over here, Anthropic is like, well, we're not going to do that. And I see a lot of companies moving toward Anthropic. This is a huge money play here. Yeah, no, I think it's interesting because like I mentioned, I think it was all about the money. And then she trying to get more users and more usage which is on the one side But on the other side I think it actually has the potential to backfire So like my experience with it which I think you make a great point is you know he says here he brings up like R movies He's like, right, like, we have different ratings for movies. Like, that's totally cool, whatever. But at the same time, you don't really have different ratings for movies in, like, a huge respect in certain movie studios, right? So, like, you're not going to expect Disney to go make R-rated movies. You know them for one thing. Now, if you have an R-rated movie studio that comes out with horror films, I will expect every single one to be R-rated. That's their brand. That's their customer base. So it's like OpenAid has to kind of choose their customer base because I felt like I had a lot of whiplash with Grok. I've talked about this in the past before. I used to be the number one Grok fan in the world. I told everyone to get it. I'm a huge free speech kind of guy. So, like, that was, like, my first, you know, take on why I wanted it. And then all of a sudden, Grok started adding in all of these, like, sexy anime avatar girls to it. And like, this is the, this is the app that I'm sitting here, like asking a question. And my son comes up to me and is like, oh, like what's, you know, when did Santa Claus die? And I'm like, look up the history of like St. Nicholas on this thing. Like, I can't have an advertisement for anime girl popping up. My wife's like, what the heck is this Annie anime thing? I'm like, my whole family turned on to it, right? Like everyone, I'm like, oh guys, get Grok. It's awesome. It's the best. All of a sudden there's all these like weird pop-ups. And so like, I just completely moved off the platform. I actually went back to open AI. So everything's chat GPT now. so it's like it's just tricky because yes i understand different people want different things for it but like you you have to set some sort of consistent boundary for me i'm like talking about you know like my family like that that tool that like i show my kids i can't hand them chat gpt if they're accidentally on my account gonna type something in and it's like well he was an adult so like we just right yeah and so um and i think like on the one side there's a family but on the other side there's also enterprise and corporate and like you you really want to make sure whatever AI platform you are using has no chance of anything in a corporate setting or a family setting of having any of that type of content. Now, if that's the type of content you're going for, there are definitely going to be platforms, you know, like an R-rated movie firm, all for that. And so I think if you try to add it on, like Sam Altman thinks like, okay, we have this huge, you know, audience nailed down. Now let's just add on this other audience. But I think that that other audience could potentially cannibalize the sales he has of his corporate. This is such a good point. I love your point about Disney. Why do people go to Disney? Because they know it's hugely guardrailed. Why do they go to, you know, whatever, sort of like one of the, go see a certain movie or something like that made by a certain production company because they know what they're getting in for. It's sort of the same with cruise. I went with Lucy on a Disney cruise. You know, she was 13 or whatever because we knew what we were getting. And then, like, kind of like across the bay, you see, like, you know, Carnival Cruise or whatever. And it is a different vibe, my friend. You know what I mean? And like, and so like, but that's, but you, but, and no, I don't mind. Like do this, you know, you're in your twenties or thirties. You want to put like, that's awesome. I did too, but that's not what I wanted. That's I think I'm, I'm wondering how they're going to figure this one out. It is absolutely fascinating. Listen, guys, this is one of these conversations that you are going to have to follow because this is going to have, you're going to have to use these things. But also as you are trying to figure out, I get asked this all the time, which model to use, which tool to use, all that kind of stuff. You have to check out AI box.ai. This is Jaden's company. This is 19 bucks a month. It's absolutely amazing. You get to sort of use all these tools side by side to compare which one is best. But now it has these incredible new features where you're actually building out new workflows, non-tech like me. I'm non-tech. It's been super, super fun for me. Guys, it's so inexpensive. I'm guessing Jaden is losing a ton of money right now, but take advantage of that. Go check out AIbox.ai, and we will see you on the next episode.