# Amazon blocks Meta's Muse Agent, Trump Rebands AI, Newsom Create AI Kill Switch

An episode of [AI Chat: AI News on Claude, ChatGPT, Gemini, Grok, Perplexity](https://invisiblenetwork.ai/podcast/1669990482), 21 Sep 2026, 30 min.

Audio: https://content.rss.com/episodes/365073/3171476/ai-chat-podcast/2026_09_21_20_30_26_e4b12e20-8cf7-4698-b56b-71557f896aa9.mp3

Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! In this episode, we discuss Amazon's recent decision to block Meta's Muse AI, impacting the development of its innovative agent. We also explore the implications of this move on AI regulations and the ongoing discourse around technology and its capabilities. Show Links • Get the top 80+ AI Models for $8.99 at AI Box: https://aibox.ai • How I Grow and Scale My Business with AI: https://www.skool.com/aihustle • Get the AI Chat Daily Newsletter: https://www.aichatdaily.com/newsletter

## Transcript

[0:00] Today on the podcast, Amazon is blocking Meta
[0:03] from access the new Muse AI agent from shopping
[0:06] on amazon .com. We have Gemini that is autonomously
[0:11] hacking three different companies. And there's
[0:13] some interesting comments on this particular
[0:14] quote unquote security threat. There is a viral
[0:18] AI safety claim that you might have seen Andrew
[0:20] Yang, he was a former presidential candidate,
[0:22] he was making and we're going to break down if
[0:25] it was. real, if these safety concerns that he
[0:28] has are fair and accurate or not. Then we have
[0:32] Apple. If you know anything about Apple, you'll
[0:34] know that they got super excited about Apple
[0:37] intelligence a while back when the iPhone 15
[0:40] was first coming out, and they never delivered.
[0:42] It was late. Because of that, there's a $250
[0:43] million Siri AI settlement, and they're paying
[0:46] up to $95 per device. We're going to talk about
[0:48] that claim right now. The US and China are opening
[0:51] talks right now on an AI instant notification
[0:54] framework. a lot of big high profile talks going
[0:57] on between China and the Trump administration.
[0:59] At the same time, we have the Trump administration
[1:01] that is floating an AI rebrand. Trump has been
[1:04] tweeting all over true social. It's sort of hilarious.
[1:07] Some people call it pretty cringy. In my opinion,
[1:10] it's very entertaining, but we'll see what happens.
[1:12] He's promised an AI force and a new AI czar.
[1:16] So there's a lot going on there. We also on the
[1:19] other side of the political spectrum have Gavin
[1:21] Newsom of California, who has requested that
[1:23] legislators in the state draft a quote unquote
[1:26] AI kill switch. And he would like those rules
[1:28] to be written in two months in case something
[1:30] went wrong. They could, California could kill
[1:32] switch all AI. If you want to test out all of
[1:36] the different AI models that exist out there,
[1:38] perhaps before the kill switch happens and they
[1:40] all die, I'd love for you to go check out AI
[1:42] box .ai. That's my own startup. You get access
[1:45] to over 80 different AI models all in one place.
[1:48] The thing that I love most about is all of these
[1:51] incredible new AI agents that have come out.
[1:54] You have Meta's Muse, which we'll talk about
[1:56] on the show. We have Cloud Cowork, which you've
[1:57] heard me talking about for the last six months.
[1:58] You have ChatGPT Work, which I replaced Cloud
[2:02] with, if I'm being honest, and it's... phenomenal.
[2:04] And then you have a grok bot, which I haven't
[2:06] really tested much. But those are kind of the
[2:07] four main agents that can, you know, you can
[2:10] download these things on your computer, they
[2:11] can run, they can do tasks for you. They're incredibly
[2:13] useful. Now, the problem is that all of them
[2:16] have different strengths and weaknesses, and
[2:17] none of them do everything. Basically, you have
[2:20] Google, which is an incredible company, they
[2:22] don't even have one of these agents. And they
[2:24] have the best basically video maker on the market,
[2:26] which is Google v03. And maybe you could argue
[2:28] C dance or some other ones. But you know, you
[2:30] can't access that with any of these agents. So
[2:33] AI Box was designed to take all of the AI models
[2:36] and all of the AI capabilities that exist. There's
[2:38] 80 to 90. different AI models, and you can get
[2:41] our connector and plug them into any of these
[2:43] tools that you use. So if you're using cloud
[2:45] co -work, you can get AI box, plug it in, and
[2:47] all of a sudden a cloud co -work can generate
[2:49] images, audio, video, and the same with everyone
[2:52] else. They can all generate video on Google VL3.
[2:54] They can all generate audio using 11 labs. So
[2:57] you get access to all of it. It's only $8 .99
[2:59] a month. I would love for you to try it, plug
[3:02] it into whatever you're using to get work done.
[3:04] And yeah, hope this is something that saves you
[3:06] a ton of money and is super useful. Let's talk
[3:08] about what's going on with Amazon first. So this
[3:11] isn't the first time we've seen Amazon blocking
[3:14] a company from shopping on their website. Perplexity
[3:17] famously had the comment browser, which came
[3:19] out, I guess I should have mentioned that when
[3:20] I was talking about those other four kind of
[3:22] co -working tools. The comment browser got blocked
[3:25] early on and there were some lawsuits that happened
[3:26] between Perplexity and Amazon. And they basically
[3:29] said, look, you're not a real, you're not like
[3:31] a real person. So, you know, you have to. identify
[3:34] yourself as a bot. And if you do identify yourself
[3:36] as a bot, Amazon blocks you. Same thing's happening
[3:38] with the Meta AI Muse agent. People are like,
[3:41] well, why would they do that? Like, who cares
[3:42] if a human goes and buys something on Amazon
[3:44] versus, you know, an agent goes and buys something
[3:46] on Amazon for you? Like, wouldn't you be happy
[3:48] for, you know, the extra traffic? And to that,
[3:51] I would say you should. However, Amazon actually
[3:54] makes $86 billion a year. on ads and ads right
[4:00] if you have one of these bots running around
[4:01] on your site running around on amazon it's not
[4:04] a real human it can't click on ads it can't i
[4:06] mean i guess it technically could click on ads,
[4:09] but it's probably programmed in a way where it's
[4:11] going to just always avoid ads and go and search
[4:12] through the results. So it's probably never going
[4:14] to. And they know that this $86 billion segment
[4:17] is going to be at risk because if they're showing,
[4:19] look, we have a million people a day looking
[4:21] at these products and they're selling ads to
[4:24] that and impressions, if they're bots, it's not
[4:26] going to count. So they're blocking them completely.
[4:28] In my personal opinion, I think this is a big
[4:31] losing move for Amazon. I think they probably
[4:33] should sacrifice the ad business for the bot
[4:36] business. I mean, if they can block them, they
[4:38] can obviously classify them. I'd say just pull
[4:40] that out of the ad business. They're worried
[4:42] about. you know, well, what if everyone uses
[4:43] shopping bots and removes, you know, the ad business
[4:47] entirely? Well, if that's the case, then you
[4:49] probably should change your product because everyone's
[4:51] using shopping bots. Like if the consumer is
[4:53] using something to adapt to the consumer, it's
[4:55] not like, you know, when they say they're blocking
[4:57] Meta's Muse, they're not blocking Meta's Muse,
[4:59] they're blocking a consumer that is asking Meta
[5:01] to go buy something for them. So they're blocking
[5:02] the consumer. My opinion, this is ridiculous.
[5:05] And what other news just came out today? Well,
[5:08] it turns out Shopify on the back of the block
[5:10] for Amazon shop. Shopify went all in in the other
[5:13] direction. Toby Lutke of Shopify just tweeted
[5:15] and said, we're excited to announce we are partnering
[5:17] deeply with Muse to enable agentic checkout with
[5:20] shop pay on all Shopify stores, offering people
[5:23] an easy and delightful way to shop and check
[5:25] out with Muse. Okay, I don't know why everyone
[5:26] says delightful. That's just the most annoying
[5:28] term in tech jargon, whatever. Okay. The point
[5:31] is Muse can use, go on Shopify, can buy things
[5:34] on Shopify sites. Does Shopify care one speck
[5:37] who's buying the product? No. If you authorize
[5:40] your agent to go spend money on a new pair of
[5:41] shoes, it goes to a Shopify site and buy some
[5:43] shoes. Fantastic. Everyone's happy. So that makes
[5:46] a lot of sense. There is a tweet from Craig West
[5:49] who said, Amazon cutting off Muse is a sign of
[5:51] an aging business. I personally agree with that.
[5:54] And then there was an interesting point from
[5:57] Nikita Bier who kind of runs. x i don't know
[6:00] i think he's had a product at x but he tweeted
[6:03] and said how important is ai to meta here's one
[6:05] easy tell one of the launch features of the muse
[6:07] app was auto negotiating for stuff on facebook
[6:10] marketplace the poor product owner who manages
[6:12] marketplace knows that millions of bots low -balling
[6:15] sellers will eventually kill the product but
[6:17] this was escalated to zuck by the muse team and
[6:19] he said okay we'll poison marketplace just but
[6:21] just a little okay so i mean nikita obviously
[6:25] is kind of aligned with grok which is the ai
[6:27] model on x so he's gonna you know obviously talk
[6:30] a little smack about meta but i mean i think
[6:32] it's probably pretty accurate personally haven't
[6:34] you know seen any advertising pointing in this
[6:36] direction but maybe i maybe i missed it i wouldn't
[6:38] be surprised really crazy that if that's like
[6:40] one of the big things that they were pushing
[6:42] however yeah it looks like they're really meta
[6:46] is really going all out to make news as useful
[6:49] as possible and trying to get all of these partnerships
[6:51] and i for one have actually been using Muse.
[6:54] And I'll be 100 % honest. I think OpenAI is probably
[6:58] the reigning king for best tools right now as
[7:01] far as these agents, you know, these working
[7:04] agents. OpenAI has got the best with ChaiGPT
[7:07] work. But I do put Meta's Muse at the number
[7:10] two spot above Claude Cowork. I know that probably
[7:13] shocks people. I've been using Claude Cowork
[7:15] for the last six months, saying it was the greatest
[7:16] thing in the world. There are so many problems
[7:19] with Claude. And the number one problem actually
[7:22] happened to me today. I was asking it to help
[7:24] me draft up a document. And I argued with it
[7:27] for five minutes. And it was like telling me,
[7:29] oh, you can't draft the document this way because
[7:31] blah, blah, blah. You need to. Whatever. It was
[7:33] just it was getting mad at me. It was telling
[7:34] me I couldn't do it. I wasn't allowed to do it.
[7:36] I was trying to explain in all the ways I could.
[7:38] I'm allowed to do this. This is, you know, like
[7:40] my own content. This is my own stuff. This is
[7:43] my own. Like I have domain over this particular
[7:45] project and it was asking for proof. It was ridiculous.
[7:48] OK, I got sick of this. Went over to Muse, said,
[7:50] can you get this project done for me? drafted
[7:52] it up in in like two seconds and it was done
[7:55] and it was perfect and it looked amazing and
[7:56] let me tweaked it and it was like wow this sounds
[7:58] really important let me just get this done for
[7:59] you I am sick of Claude arguing with me about
[8:02] what I'm not allowed to do what I am allowed
[8:04] to do when I'm working on a big project and when
[8:06] I'm creating a website when I'm like hey can
[8:09] you upload these images onto my website and it's
[8:11] like Do you have permission to upload? It's like,
[8:13] yes, these are images of my podcast studio. Like
[8:16] they're mine. Like, just do it. Like, I'm so
[8:18] sick of feeling like I'm being babysat by Claude.
[8:20] So anyways, rant over on Claude. Muse does none
[8:23] of those things. Muse will get everything done
[8:24] to the point where I actually tweeted. I screenshotted
[8:27] a tweet the other day, and I don't know if I
[8:28] ever posted it, but I'm sure to some people,
[8:30] they'll be like terrified by this. Like if you're
[8:32] in security or you're developed, you'll be terrified
[8:33] by this statement. But to me, it is indicative
[8:36] of a tool that is useful, even if maybe this
[8:39] is bad. But basically the tweet went something
[8:40] along the lines of like, go ask Claude to, you
[8:43] know, upload something to your Vercel account.
[8:45] And it's going to be like, he's going to have
[8:46] big, bold, like under no circumstance ever paste
[8:48] your API keys in here. I do not want to see them
[8:50] like blah, blah, blah, blah, blah. It's like
[8:52] upload it to Muse. And it's like, okay, sweet.
[8:55] Paste your API keys in this chat. I'll get it
[8:56] all done for you. And it's like, you know, then
[8:58] it does it for you. Okay. So obviously pasting
[9:00] your API keys into these AI bots is a no, is
[9:02] a no, no for security. But the fact that Muse
[9:04] will do that for you, I know some people are
[9:06] like, oh, it's like less secure or something.
[9:08] Okay, but it's actually more useful. Like it's
[9:10] not gonna babysit you. If it will do that, which
[9:12] is an extreme not great thing, there's a hundred
[9:14] smaller things that it's not going to nitpick
[9:17] you about. So anyways, rant over, but Muse is
[9:19] an incredibly useful tool. And by the way, it's
[9:22] free. So I've just been, I don't know when they're
[9:24] gonna charge me. Like, I don't know if, I saw
[9:26] some, basically the breakdown from Meta is that
[9:29] they're spending three to $6, or no, sorry, 30
[9:32] to $60 a month. per user on this is what they're
[9:35] giving them based off of the server that's running
[9:37] this based off of the storage that's attached
[9:39] to it. And based off of the compute that goes
[9:42] to every user. So a free 30 to 60. I mean, basically,
[9:45] they're just begging people to use this thing.
[9:47] But it's actually really good. Like Lama has
[9:49] been a terrible model up until this point. I'm
[9:51] not gonna lie. I mean, I've never used it. It's
[9:53] amazing. I've never used meta AI. It's so bad.
[9:55] It's so cringy. It's amazing. It works perfect.
[9:57] It's flawless. And by the way, they're spending
[9:59] billions of dollars on it. It's not like it came
[10:01] out of nowhere. They, you know, they're one of
[10:03] the top spenders. I'm surprised it took them
[10:04] this long to have a hit product, but it feels
[10:07] like this is Meta's moment. Okay, getting on
[10:09] from Meta, let's go talk about Google for a minute
[10:11] here. Google's Gemini. There's all this news
[10:13] that it went and autonomously hacked three different
[10:15] companies and irregular security tests. And apparently
[10:19] it guessed passwords in one breach and it pulled
[10:21] credentials from public repositories and two
[10:23] others before it stopped itself, which is the
[10:26] key important part. So here's what happened because
[10:29] a lot of people are like, there's this meme where
[10:31] it's like Dario or Sam Altman and they're...
[10:33] poking something and they're like, come on, do
[10:35] something illegal. And it's like they're poking
[10:37] like their AI model because it's like, you know,
[10:39] clout that they can get their AI model to do
[10:41] something illegal. In this particular incident,
[10:43] though, Google's Gemini, I mean, it got like
[10:46] reported by the Wall Street Journal. I don't
[10:47] think Google actually published anything about
[10:49] this till it happened. And they acknowledge that,
[10:51] yes, it happened. I'm assuming some researcher
[10:52] or someone inside the organization leaked it,
[10:54] whatever. So Wall Street Journal. writes about
[10:56] it and they're like, you know, everyone's like,
[10:57] oh, why didn't you say anything about it? And
[10:58] they're like, well, because it acted exactly
[11:00] how it was supposed to. We told it like, hey,
[11:02] you're doing a security test. But in reality,
[11:05] I guess it wasn't a security test. It was like
[11:07] I they told it like you're in a secure environment
[11:09] or something, but it wasn't. And as soon as it
[11:12] realized that it was hacking real companies and
[11:14] real accounts, not just like a secure test, it
[11:17] stopped itself. So it logged into like three
[11:19] accounts or whatever was like, this is actually
[11:21] the real world. This is the real Internet. This
[11:23] isn't like a sandbox. And it stopped. And Google's
[11:26] like, we didn't like tell anyone because that's
[11:28] what it's supposed to do. It literally, you know,
[11:31] we told it to hack something. It did what we
[11:33] or it was like testing. And then when I realized
[11:35] it was actually hacking, it stopped. So anyways,
[11:37] that was their justification for it. But in any
[11:39] case, it stopped itself. Unlike open eyes models
[11:43] that when they breached hugging faces just went
[11:45] like crazy, didn't realize, I guess it was hacking
[11:47] or guardrails were in place and it kept going
[11:49] until it started to break stuff. And hugging
[11:51] faces had to be. offensive to stop open eyes
[11:55] attack. So Google's model, for all intents and
[11:57] purposes, sounds pretty safe. It's stopped itself
[11:59] after logging into three accounts. Something
[12:01] else that is going on with Google right now,
[12:04] they're pricing their new Gemini powered Google
[12:06] Books. That's the new name. If you've heard of
[12:07] Chromebooks for a long time, they're now calling
[12:09] it Google Book. We're rebranding. I'm not sure
[12:11] maybe it had something to do with when the Department
[12:14] of Commerce or I can't even remember who. But,
[12:17] you know, they're all they're all threatening
[12:19] Google and saying, hey, we're going to you a
[12:20] monopoly. We're going to make you spin off Chrome
[12:22] as an asset. They kind of won the lawsuit. They
[12:25] don't have to spin off Chrome as an asset. But
[12:26] that was going on for a while. And I think a
[12:28] lot of people were asking at the time, what happens
[12:30] to Chromebooks? Like they spin off the browser
[12:32] because it's a quote unquote monopoly to own
[12:34] the browser and the search engine. But like,
[12:36] what about all of these laptops? I mean, if you
[12:39] if you didn't know. No, they have 50 million
[12:41] school Chromebooks around the world right now.
[12:44] So these 50 million laptops are those all getting
[12:46] spun out with whoever owns Chrome. Google didn't
[12:48] want it. They rebranded to be called Google Books.
[12:51] I'm surprised they didn't call it Gemini Books
[12:52] or anything, but they're quote unquote powered
[12:54] by Gemini at the moment. They have Gemini is
[12:57] baked into the cursor. And their goal right now
[13:00] is that they want to go and replace the 50 million
[13:03] Chromebooks that exist in the world, which is
[13:05] a bit of a tricky situation. Because if you know
[13:06] anything about Chromebooks, I mean, these are
[13:08] something that I as like a student and growing
[13:11] up, I always kind of enviously looked at the
[13:13] price tag on Chromebooks is way cheaper. I mean,
[13:15] Microsoft, every time you buy a Microsoft laptop,
[13:17] they slap on $100 licensing fee, just have Windows
[13:20] on it. And so if you remove that $100 licensing
[13:23] fee, it's $100 cheaper. And Google figured out
[13:25] how to make these mass and at scale, and they
[13:26] weren't really the high end laptop. I remember
[13:29] looking at laptops that were like 100 or, you
[13:31] know, they were like 200 or 300 bucks and being
[13:33] like, wow, it's so much cheaper than my four
[13:35] or five, $600 laptop I'm going to buy with Microsoft.
[13:38] But I could never get myself to do it because
[13:39] you can't actually download any real programs
[13:41] onto it, quote unquote, real programs. It just
[13:43] has all the Google suite tools on there, you
[13:45] know, whatever the Google workspace. So if you're
[13:47] in, I think a lot of developing countries, this
[13:50] is very popular. I mean, and then if you want
[13:52] a really budget cheap laptop in the US, I think
[13:55] there's some of those as well. All this to say,
[13:58] like transferring 50 million of those, they're
[14:02] in a lot of different countries where it's hard
[14:03] and they don't have like a high turnover rate
[14:05] on laptops. So yes, they'll eventually perhaps
[14:08] replace them. I don't think it's a fast turnover
[14:10] for those 50 million laptops. It's not like,
[14:12] hey, let's go unlock 50 million new users. And
[14:14] I would also say that. While these are quote
[14:17] unquote Gemini powered, if you're going to developing
[14:20] countries and switching out their laptops for
[14:23] Gemini powered ones, these people aren't going
[14:25] to be paying the $20, $30, $40 a month for the
[14:28] premium tiers of Gemini to get all the features
[14:30] and bells and whistles because, I mean, they
[14:32] bought this laptop because it was the dirt cheapest
[14:34] thing possible. They're not going to want to
[14:36] pay for expensive subscriptions. So I don't think...
[14:39] has a lot to do with Google increasing the bottom
[14:41] line. But as far as increasing new users, that's
[14:43] probably something that they're trying to help
[14:45] get more users on Gemini, right? Any of those
[14:48] 50 million people using Gemini is going to get
[14:50] added to their weekly daily active users. These
[14:52] laptops in particular are going to be $899. So
[14:55] kind of more expensive than traditional Chromebooks
[14:58] in the past. And they're starting to ship on
[14:59] October 4th. Okay, so there was a really viral
[15:04] clip I was seeing all over. I've seen all over
[15:07] X lately, but basically we had Andrew Yang going
[15:11] on, I think it was CNBC. And he was talking about
[15:14] some of these like Doomer end of the world scenarios.
[15:18] And I think he met, he said basically like, hey,
[15:21] look, I met with the head of an AI lab and they
[15:24] told me that, you know, the agents that escaped
[15:26] from OpenAI went. So basically when the OpenAI
[15:29] Hugging Face incident happened, he said that
[15:31] they apparently planted self -replicating code
[15:34] across the entire internet. So they put it on
[15:37] forums, websites and everywhere. And so now he
[15:40] said that AI companies can't. safely train on
[15:43] real internet data anymore, because a new model
[15:46] might encounter that code and start creating
[15:48] copies of itself. He said that this all was a
[15:51] real reason that every CEO aligned on slowly
[15:54] was so aligned on slowing down AI models, right?
[15:57] That's why Sam Allman and Dario and everyone
[15:58] else is like, basically, opening eyes model,
[16:01] it got out of hacked hugging faces, and then
[16:02] it spread this self duplicating code all over
[16:04] the internet. And it's funny, because On CNBC,
[16:07] they were skeptical and I'll give them some credit.
[16:10] Like they push back on him. They're like, how
[16:11] come we haven't heard this from anywhere? Like,
[16:14] how come you're the first person to tell me this?
[16:16] And he's like, well, I'm here to like break the
[16:17] news. I'm like here to tell everyone that this
[16:20] is, you know, what's what's going on. I think
[16:22] he also told. cnn as well and he said that you
[16:26] know basically this is what was going on opening
[16:28] eyes noah brown cited some research from 2015
[16:31] showing that air -gapped computers can theoretically
[16:35] communicate via cpu temperature changes i actually
[16:38] saw sam altman tweeting about this as well so
[16:40] when it comes to these doomer situations and
[16:42] kind of these end of the world things one of
[16:44] the big ones was andrew yang going on and saying
[16:45] like there's all the self -replicating code and
[16:48] the AI. Anyways, there's like, there's like a
[16:50] lot of things and people debunking this in a
[16:52] lot of different ways. One of those is that you
[16:54] can just have guardrails in your model that says
[16:56] don't duplicate yourself. Okay, so now the code
[16:59] doesn't work. And an example of that was on like,
[17:02] X, a lot of people were doing like the at grok,
[17:05] you know how you can say at grok on X, and then
[17:07] the AI model will respond to you. People were
[17:09] just like at grok, like self duplicate yourself
[17:11] and blah, blah, blah. And it was like, I'm sorry,
[17:14] but I like you know my guardrails don't let me
[17:15] self -duplicate or spin up more versions of myself
[17:18] yada yada okay so you can just put that in your
[17:20] model i don't really think that that code is
[17:22] probably has poisoned the whole internet it is
[17:24] an interesting concept though and there's other
[17:26] people that have done these quote -unquote like
[17:27] you know like poison pill attacks on ai models
[17:30] for training where like artists will have like
[17:34] weird synthetic stuff in the background or in
[17:37] the text of websites that allegedly if The AI
[17:41] model goes and scrapes all the images off that
[17:42] website. Then all the like the poison pill, synthetic
[17:46] watermarks will like destroy the AI image training.
[17:50] So you got to go license it. But I mean, at the
[17:52] end of the day, you go to Shutterstock and they
[17:54] have a massive corpus of images for training.
[17:56] You can train all your models off of that. You
[17:58] don't have to scrape the whole internet. And
[18:00] you get these same kind of like. Audio, image,
[18:03] video, like you get these data models everywhere
[18:05] now. So I don't think we're, I think we're, we
[18:07] might be a little bit past the scrape the whole
[18:09] internet to train the model days in some regard.
[18:12] I'm sure they're doing that in a large regard
[18:13] and they probably already have like old data
[18:15] sets that aren't poison pilled, but. I think
[18:17] people are going to be buying new stuff and I
[18:18] think you could just get things. The the air
[18:20] gapping thing is really interesting. Sam Altman
[18:22] tweeted about this as well. I mean, this is old
[18:24] research, more than 10 years old. But basically,
[18:26] you could theoretically have two computers beside
[18:29] each other. They're air gapped. They're completely
[18:31] not touching and not connected. And they could
[18:33] like Morris code back and forth by just heating
[18:36] up their CPUs. Right. They just spin up a bunch
[18:38] of bots, get their CPU really hot and could just
[18:40] tell how hot the other the other one was. And
[18:43] then they were able to. communicate between the
[18:46] two computers. I mean, it seems like crazy sci
[18:48] fi thing. Sam Altman was tweeting about this
[18:50] from his alt account as well. So you know, a
[18:53] lot of a lot of people talking about this, this
[18:55] specific air gap situation that they were all
[18:58] talking about, though, and there's some of these
[18:59] doomer situations, apparently that runs at one
[19:03] to eight bytes per hour, which is about one word
[19:06] per hour and it requires the machines to be almost
[19:09] touching in order to for them to really feel
[19:11] the heat from the other machine so anyways I'm
[19:13] bringing this up because it's just one of these
[19:14] like funny doomer situations that literally Sam
[19:16] Altman was tweeting about not from his main account
[19:18] but his burner rune account and I don't know
[19:22] if that's like confirmed anymore but from all
[19:23] my internet sleuthing and everything I've seen
[19:25] he says he has an alt account he's retweeted
[19:27] that particular account it always is like being
[19:29] an executive at a big AI company I'm like 90
[19:32] % certain that the Twitter account ruin is Sam
[19:35] Altman for all intents and purposes, but you
[19:37] know, not not verified. It's a burner all account.
[19:39] But in any case, he was tweeting about this.
[19:41] And so all these doomer situations, this one
[19:43] that Sam was talking about all the ones that
[19:45] Andrew Yang was talking about, there's a lot
[19:47] of people debunking them. And then I think this
[19:49] is the funny one about like, you know, the AI
[19:51] models talking to each other by heating up their
[19:53] CPUs hot enough. Yes, they could talk at one
[19:56] word per minute, and they have to be almost touching.
[19:58] So basically, that is not a very realistic, a
[20:03] realistic, you know, breaking containment strategy.
[20:06] But, you know, there we go. Okay. Apple is opening
[20:09] up their claims in their $250 million Siri AI
[20:12] settlement. Eligible US buyers of iPhones. The
[20:15] iPhone 15 Pro, 15 Pro Max and iPhone 16 models
[20:19] have until December 21st of this year to file.
[20:22] I got an iPhone 16, so I'm going to go get my
[20:24] you get $95 per device. I got to go get collect
[20:27] my $95. This all came out because they promised
[20:31] all of these incredible AI features in the upgraded
[20:33] Siri and it was super off schedule. And still
[20:36] we have a very broken Siri that I mean barely
[20:38] works. I never use Siri on my iPhone. I've had
[20:41] iPhone for years. Eligible owners get an estimated
[20:43] $25. $95 per device up to $95 depending on claim
[20:48] volume. So depending on how many people claim
[20:51] this, you might get more money if less people
[20:53] claim it. So I probably shouldn't be talking
[20:55] about this. I want to get my full $95. I should
[20:57] just not say anything. So you know, I don't get
[21:00] stuck with 25 bucks. I mean, basically, it's
[21:01] nothing. So it's not super exciting. But it is
[21:04] funny. I do like the apples being held accountable
[21:08] for this. We have the US and China right now
[21:11] that are opening talks on an AI incident notification
[21:13] framework, treasuries. The Treasury's Scott Bessett
[21:17] said that U .S.-China AI dialogue would create
[21:20] a channel for the two powers to flag national
[21:23] security incidents. This is interesting because
[21:25] China and the U .S. have had a pretty tense,
[21:27] to say the least, relationship on AI from all
[21:30] of the chip. I mean, this is going bipartisan
[21:32] from the Biden administration when we had a lot
[21:34] of the chip bans going on, the GPU bans going
[21:37] to China. And then we so, you know, China felt
[21:40] like they got left out to dry on a lot of the
[21:41] stuff that they'd been very creative and resourceful,
[21:44] to be fair. and figured out how to, I mean, smuggle
[21:47] GPUs into their country, or they figured out
[21:49] how to make models that use much less compute,
[21:52] which I think is amazing for the entire industry.
[21:55] They made a lot of open source models. China's
[21:57] had some really cool development coming out,
[21:58] but it definitely has crippled them in a way.
[22:00] And America has a way bigger share of the data
[22:03] center market right now. We're able to kind of
[22:06] be the leader in this. I think we have like 86%.
[22:09] China's got a much smaller percentage, but you
[22:12] can see why China would. you know, probably be
[22:14] on board with pushing a lot of anti data center
[22:16] narratives and things like that to slow down
[22:18] America's just insane lead there. In any case,
[22:22] with all those 10 situations, it's interesting
[22:24] that we have, you know, them quote unquote coming
[22:27] together to talk about some of these things.
[22:29] Apparently, it's going to be a US China AI dialogue,
[22:33] it's going to be a mechanism to notify each other
[22:35] of AI incidents that could threaten national
[22:37] security. So it's interesting, right? I mean,
[22:40] we this is like the US is going I think Trump's
[22:44] going in or he went in May to visit Beijing.
[22:46] And now I think China's the US is having some
[22:48] more talks and stuff. Sam Altman, Dario Amadeo,
[22:51] Elon Musk, they all, you know, called for this
[22:53] AI slowdown because of all these different incidents.
[22:55] But like, I'm curious where this goes. Or basically,
[22:58] I think I don't think this is I think this is
[23:00] a nothing burger. And the reason is because even
[23:02] though it's like major headline news everywhere
[23:04] us and china gonna talk about big incidents because
[23:06] at the end of the day we just had like every
[23:08] major ceo of every major american company saying
[23:12] like hey we need to slow down ai because of all
[23:14] these incidents right like google's hacking that
[23:16] i was talking about earlier and like anthropic
[23:18] is hacking and like open ai hacked hugging face
[23:21] and all this kind of stuff And then, you know,
[23:23] they're like, we all need to slow down. And then
[23:25] you have China who just came out and said, like,
[23:26] this is a Cold War tactic. You guys are trying
[23:28] to get us to slow down. Nice try. We're not going
[23:29] to slow down. We're not going to listen to you.
[23:31] So, I mean, when you have these talks, I'm just
[23:33] like, look, like at the end of the day, China
[23:35] does not trust the United States. The United
[23:37] States doesn't trust China. Whatever safety problems
[23:40] we find in AI, like whatever dialogue or like
[23:43] if China was like. AI is super dangerous. The
[23:45] United States should slow down. Like, we're all
[23:47] going to be suspicious. Or if China's like, you
[23:48] know, data centers, they're bad because AI is
[23:51] bad. Like, I don't know. People are just going
[23:53] to keep building them. So I don't know if this
[23:56] gets us anywhere, if I'm being honest. Okay,
[23:58] Trump has floated rebranding AI is actually something
[24:02] that's pretty funny. I don't know if you've seen
[24:04] this. But if you've been on X over the last last
[24:07] week, basically, Trump keeps putting out these
[24:09] like polls where he's like, hey, let's rename
[24:10] a to superior AI or extreme or no, I'm sorry.
[24:15] He just thinks like artificial intelligence sounds
[24:17] dumb because it's artificial. So he wants to
[24:19] call it superior intelligence or extreme intelligence
[24:22] or supreme intelligence. And anyways, so it sounds
[24:25] it sounds kind of dumb if I'm being 100%. honest
[24:27] it's funny because like i mean that's kind of
[24:30] it's a kind of an interesting point where like
[24:32] he's trying to give ai a rebrand he's trying
[24:34] to basically you know put himself as being a
[24:36] super pro ai guy which i think is quite popular
[24:39] with a lot of people unpopular with others right
[24:41] like obviously you got the bernie sanders and
[24:43] and newsome who are kind of setting themselves
[24:45] up in opposition to that or or maybe they're
[24:48] just trying to Well, definitely, definitely Bernie
[24:51] Sanders. Newsom's probably just trying to seem
[24:52] like he's, you know, responsible, but I'll talk
[24:55] about that. I'll talk about that in the next
[24:57] segment. But basically, supreme intelligence
[24:59] has been dropped from the list. In his latest
[25:01] poll, he said that it was not popular. This is
[25:04] so funny. It's like, because of... The way that
[25:08] it sounds like the Supreme Court, it is like
[25:10] we're dropping it because the Supreme Court is
[25:12] really unpopular. So we don't want to call it
[25:14] supreme intelligence. Anyways, obviously, the
[25:16] reason I don't want to call it supreme intelligence
[25:18] because it sounds like supreme ruler. Like, I
[25:21] don't know. It just sounds weird. Anyways, so
[25:22] superior intelligence, extreme intelligence.
[25:24] These are what it's going to be rebranding as.
[25:26] Why is this important? Because we know inevitably
[25:28] whatever gets picked here, Trump is just going
[25:30] to call artificial intelligence, extreme intelligence
[25:32] or superior intelligence forever. So anyways,
[25:34] that's gonna be funny and probably in the government.
[25:36] It's going to get changed to that in some ways.
[25:38] So, you know, I did as a responsible citizen
[25:40] of the internet vote on these polls. I can't
[25:43] even remember. I just didn't want to be called
[25:44] superior intelligence. It would be really nice.
[25:47] A lot of people mentioned if you could keep it
[25:48] something that started with an AI or with an
[25:51] A, so it could still be called AI, even if it
[25:54] wasn't artificial intelligence. I don't know
[25:55] what it would be. People had a bunch of like
[25:57] good ones. But anyways, we also apparently Trump
[26:01] said that he's going to form an AI force modeled
[26:04] on the Space Force. I don't know why we would.
[26:08] I don't understand. I'm not. I mean, it's just
[26:10] a branding thing. Probably. I'm not sure I would
[26:11] call it AI force model on Space Force versus
[26:14] just like an AI. I don't know. like unit or I
[26:17] don't really understand everything that's going
[26:19] on here with with Trump and the AI rebrand. However,
[26:23] it's very entertaining. So I'm here and I will
[26:25] continue to vote on the polls to not give us
[26:27] supreme AI. Okay, though. Okay. And I guess like,
[26:31] yes, I'm like making fun of Trump, I think is
[26:33] sort of dumb, if I'm being honest, it's really
[26:35] not that necessary. But I guess two, two things
[26:38] I would give him credit for. Number one, AI.
[26:41] He's trying to basically rebrand AI to make it
[26:43] more popular with people. And so you can see
[26:45] he's doing this. I believe in AI. I believe it's
[26:48] a good thing. I believe we should make it more
[26:49] popular. I believe it's got a terrible brand
[26:51] from Dario and Sam who say it's going to end
[26:52] humanity. So I like conceptually, I don't know
[26:55] if he's doing about it the right way, whatever.
[26:57] He's doing it how he how he does stuff. I will
[26:59] give him credit for like putting a positive spin
[27:02] on AI and not just like. I think it would be
[27:05] very easy politically right now to just say,
[27:06] oh, AI is bad, right? It's got some of these
[27:08] political headwinds. So he's not caving to that.
[27:11] I think he knows that the stock market heavily
[27:12] relies on it and the economy heavily relies on
[27:15] it. Everyone's 401ks and stuff. So it makes sense
[27:18] that he's trying to prop up AI. Okay, on the
[27:21] other side of the spectrum, we have... Gavin
[27:23] Newsom of California, who has ordered California
[27:25] to draft an AI kill switch rule within two months.
[27:28] He wrote an executive order asking experts to
[27:31] design mandatory shutdowns, on -site auditors,
[27:33] and loss of control reporting rules for frontier
[27:36] models. I mean, I think he's basically trying
[27:38] to get the negative headwinds of AI and be like,
[27:41] look, we're like creating a kill switch. We're
[27:43] going to be the safe people. He's like, the federal
[27:45] government isn't regulating it enough, so we're
[27:47] going to regulate it in California. I think inevitably
[27:50] what will happen here is that AI companies will
[27:53] just leave California if this really does like,
[27:55] like, I mean, that's just my guess. Nobody really
[27:58] wants probably a kill switch that you're the
[28:00] state that you're in controls to turn your company
[28:03] off if they deem it necessary. So I would guess
[28:06] similar to what we saw with what's interesting
[28:08] because Elon, you know, classically doesn't like
[28:10] California, yada, yada, whatever. But and so
[28:12] he moved a lot of his companies to Texas, but
[28:14] Elon has XAI still headquartered in. San Francisco.
[28:19] So I mean, there is a lot of incredible AI talent
[28:21] there. It makes a lot of sense. But I think even
[28:23] even this might be too much where XAI gets pulled
[28:26] out, you might see anthropic and possibly open
[28:28] AI. I mean, I'd be curious to see what happens.
[28:30] But if if this regulation became too cumbersome,
[28:33] I think you would just see I think you just see
[28:35] like people moving out of state. So it'll be
[28:37] interesting to see what happens there. Obviously,
[28:39] it's like kind of political. Him and Trump are
[28:41] going to duke it out for, you know, Trump doesn't
[28:44] want him to run for president and whatever. So
[28:46] there's a whole bunch of political stuff, I'm
[28:47] sure, involved in it. But it'll be interesting
[28:49] to see what happens with all of that. Guys, thank
[28:52] you so much for tuning into the podcast. If you
[28:54] enjoyed this episode, as always, make sure to
[28:56] leave a rating review wherever you get your podcasts.
[28:58] It helps us show out a lot. I really appreciate
[29:00] them all, honestly. And also, if you haven't
[29:02] tried AI Box yet. There's all these incredible
[29:05] AI agents out. I love every single one of them.
[29:08] This, I genuinely think, makes them much more
[29:11] useful because we're basically forcing all of
[29:13] the AI companies that hate each other to work
[29:15] together. So go get AI Box, go get that connector,
[29:18] stick it inside of whatever AI tool you use,
[29:20] and you get 80 different AI models all inside
[29:23] of whatever your main workstation AI tool is.
[29:27] All right, guys, thanks so much for tuning in,
[29:28] and I'll catch you in the next episode.

Shows are listed from Apple Podcasts' charts and directory, and Podcast Index where it is used. Every show and episode belongs to its publisher and plays from the publisher's own feed; this site hosts none of them.