← Everyday AI Podcast – An AI and ChatGPT Podcast
Everyday AI Podcast – An AI and ChatGPT Podcast
Ep 704: AI bots start their own social network, NVIDIA-OpenAI drama, Google’s huge releases and more AI News That Matters
00:41:06
Loading summary
Transcript13 lines
- [00:01]
Jordan Molson
This is the Everyday AI show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business and everyday life.
- [00:18]
Modulate Representative
Most AI tools that analyze sales or customer support calls turn those conversations into a text based transcript. But text only transcripts miss most of the value.
- [00:29]
AI News Host
And Modulate fixes that.
- [00:31]
Modulate Representative
Modulate's new Velma voice native AI model and their ELM technology actually understand what's.
- [00:38]
AI News Host
Happening on those calls.
- [00:39]
Modulate Representative
It picks up the valuable tone, timing, emotion and intent that all AI transcription tools can't provide. So whether it's for sales, customer support, or voice agents, Modulate's new Velma model helps you capitalize on what text only AI tools miss. Demand more from your AI today with Modulate Modulate at Modulate AI.
- [01:02]
AI News Host
All right, so I know sometimes that I say it's been a wild week in AI or a weird week, but this week, I mean, AI agents started their own social network, no humans involved, and it's going bonkers. There's a new Wall Street Journal report that said the hundred billion dollar deal between Nvidia and OpenAI is on ice. And Anthropic CEO penned an essay that's like the length of an encyclopedia that said that could very well take half of white collar jobs and maybe kill millions. Yikes. Oh, and between all those juicy AI news stories, Google ships some of the most useful AI updates that are getting no headlines at all. Like a Chrome that can browse for you and this genie model that can simulate entire worlds instantly. So yeah, this week, an actual wild and weird week. All right, so if you missed anything, don't worry. This is our everyday AI episode where we go over the news of the week called AI News that Matters. So welcome. And what's going on? My name is Jordan Molson and welcome to Everyday AI. So if you're new here, this is your daily livestream podcast and free daily newsletter helping everyday business leaders like you and me keep up with all the AI craziness, how to make sense of it, and to grow our companies and our career. So if that's what you're trying to do. Suite starts here with the unedited unscripted live stream podcast. But if you want to be the smartest person in AI around, you look around. Do you want to be the smartest? Well, go to our website, your everyday AI.com. it's a cheat code. So go sign up for our free daily newsletter. Each day we recap the daily livestream podcast and all the other AI news and you can go listen to 700 episodes for free from some of the smartest minds in the world. All right, so enough chit chat. Let's get to the AI news that matters. Like I said, we do this almost every single Monday. Recapping the AI news. No fluff. We just tell you what's going on and what's going on this week. A bunch of wildness. So let's start with that hundred billion dollar deal. Is it on ice? Well, maybe, maybe not. So Nvidia CEO Jensen Huang told reporters in Taipei over the weekend that Nvidia will make a very large investment in OpenAI potential its biggest ever, reaffirming a major cross industry push and clarifying some recent reports that said otherwise. So Nvidia does plan to make a huge investment in OpenAI, Jensen said on Saturday in Taipei and called OpenAI one of the most consequential companies of our time. So this comes though, amid a recent blockbuster report from the Wall Street Journal that said the $100 billion circular financing deal kind of right between Nvidia and Open, quote unquote, on ice. But Wong confirmed that Nvidia will absolutely be involved in the next financing round for OpenAI that CEO Sam Altman is closing, though the final amount, he said is for Altman to announce. So this comment follows Nvidia's September announcement that it could invest up to $100 billion in OpenAI. Though Wong did say that that original figure was non binding in not finalized. So Wong denied some of the Wall Street Journal and other reports that said he was unhappy with OpenAI and called suggestions to that effect nonsense, while also having privately expressed concerns about OpenAI's business discipline and competitive threats, according to other reports. So he did emphasize the company's close working relationship. And Sing singled out Sam Altman my name, saying that he really loves working with Sam and that Open AI is incredible. So pretty wild turn of events here in the last, I don't know, 48, 72 hours. We saw reports, you know, I think late Friday from the Wall Street Journal that said, hey, Nvidia and OpenAI not looking so good. That hundred billion dollar deal is on ice. And then Wong essentially said, nonsense. He essentially said it's happening, but it's up to, you know, Sam Altman and OpenAI how much that amount is speaking of that mount. Yeah. More news here in a couple of minutes on a $60 billion commitment from a couple of the players in big tech. But our next piece of AI news, something that didn't get a lot of love, but it's definitely worth shouting out, especially if you are into watching really cool things made by AI. So Google is making its Genie 3 environment generation model available to outside users after they demoed it themselves. Intern. Now the whole world, well, some of the world can go use it via Project genie, offering a first look at how AI can build explorable simulated worlds. So unfortunately, right now it does require Google's AI Ultra plan, which is $250 per month. Also, if you want to use it, you have to be a US resident aged 18 or older. So it's not out for everyone. But for the people that do have it, I think it's pretty amazing, right? I mean, if you've seen any, like, let me just say in straight, unscripted terms what this is, I mean, number one is bonkers. Number two, it's amazing to watch. So you can essentially simulate a real interactive world. So think of, you know, pretty good video game graphics. I don't play video games and I haven't forgotten, I don't know, 20 years maybe, but it looks like, I don't know, like a video game from five years ago. You can take a photo of yourself, put a prompt in there, and it looks like a video game version of yourself doing whatever you said. It's literally quite amazing. And it renders it in real time. That's the crazy thing. These GENIE worlds in Project GENIE are explorable by the user, right? So it's not like a video that you watch. They are interactive and they render in real time. Which, again, the fact that AI technology is at this point and it looks this good is nano bananas. So at launch, Project Genie includes three interaction modes. World sketching, exploration and remixing. Also, it uses Google's Nano Banana Pro image model to produce the initial Source image that Genie 3 then converts into that exploration explorable world. So users can define character description and camera perspective and can write custom prompts for worlds other people have already created. So right now, generations are deliberately constrained though. So each generation runs for up to 60 seconds and playback is capped at 24 frames per second at 720p. So it's not full HD or 4K or anything like that, but who, who freaking cares? It's, I mean, some of the coolest, you know, AI technology that you can play with. Since I don't know anything. I mean, sure, ChatGPT was fun and all, but this is. Doesn't make sense that something this good exists. Yeah, you got to be on that extra pricey Google Ultra plan. Gemini Ultra plan and in the US to use it right now. So I think a lot of the talk on this has been around kind of like video game, but I don't think that's what this is. Right. Anything that's gone, you know, semi viral online or has gotten a lot of news and media exposure is essentially people, you know, inserting themselves into a, you know, Mario esque world. And it's like, okay, here I am in a Chicago version of, you know, Mario for 60 seconds and I can interact, you know, I can move myself. So it's like, okay, that's, that's cool, right? And that's what's getting a lot of the headlines right now. But that's not the end game here, people. So this is really for, well, two things. It's for agent understanding, number one and number two, this is for Google to get more data of how people interact with the real world. Right? Because the more people that use this, the more data that Google gets and they understand how people interact with all these elements and it is extremely realistic. Right. And there's different, you know, maneuvers you can do, you can, you know, jump around things. And I've seen, you know, examples where you can have a hang glider going from point A to point B. But the physics are really, really good. Right. This isn't the, you know, Will Smith eating spaghetti, hand going through the neck. It's extremely realistic video game realistic. Right. But this is ultimately, I think so Google can get it even bigger leg up on their lead in multimodal AI, but specifically to help Gemini Robotics. Right. I think that's, yes, there's agentic plays here, but I think more than anything this is so Google can get more data with how AI interacts in the real world, right? Yes. We can talk about scaling laws and you know, reinforcement learning and all of those things when it comes to text based AI. But the AI that I think is ultimately going to make the most economic impact, especially in the2030s is, you know, embodied AI, right. Maybe even in the, you know, later part of the 2000s here in maybe two or three years, that's what it's going to be. But we need more data, more training data and I think if nothing else, this gives Google a huge leg up. All right, our next piece of AI news, two big announcements from Anthropic that.
- [10:47]
Modulate Representative
I think Anthropic is finally starting to.
- [10:50]
AI News Host
Put some non developer love out there for the general business community. So two new things. So Anthropic has rolled out interactive apps for Claude that embed simplified app experiences as well as new plugins. So first the new interactive apps use Anthropic's model context protocol to share structured content between Claude and third party apps, enabling end to end workflows such as, you know, reading your email, enriching company data and generating presentations that you can see and interact with all in the same chat window without having to leave or copy and paste anything. So right now it is only available to people on paid Claude plans on the desktop or web for now. And some of the first apps that are included in the first rollout include Canva, Clay, Gamma Amplitude, Asana Box, Figma, Slack and Monday.com so those are just the apps, right? So there's kind of a difference between the apps in the connectors. So you know, as an example, they have a connector for Gmail and for some Microsoft products, right. So those essentially can read data, but the apps can generate interfaces. Right. So a simplified version of some of these websites are going to be in the in your Claude chat window. Right. And the ability to write as well. So you know, pretty big difference between connectors and what's been possible inside of Claude versus what's now possible with these interactive apps separately. Anthropic has launched plugin support for Claude Cowork. So that lets Claude act as a specialized assistant for different departments. So how this works, the plugins bundle skills, data, connectors, commands and sub agents. So for example, a sales plugin could hook Claude into a company CRM and knowledge base and add commands for customer research and call follow up. So Anthropic has also open sourced 11 plugins spanning productivity, data analysis, marketing and customer service. So pretty big update here for Claude Cowork. So if you don't know what Claude Cowork is, this is the more, you know, business friendly, right? I'll say non technical version of Claude code. So it's essentially a version of Claude code with a graphical interface, you know, that's more drag and drop friendly versus you know, running something out of a command line, you know, terminal tool. So if you want more information, in episode 707 last week we did cover the new CL Claude apps. And then in 696, episode 696 we covered Claude Cowork. So not the plugins that are now available for Claude Cowork, but we did cover Claude Cowork. So make sure you go check those episodes out. All right. But maybe one of the most useful AI updates that we get, we all get right now or by we all the most people. Well that's if you use Google Chrome and you have a paid Google account, you are in luck. It is Christmas in February for those people. So Google has rolled out some new major AI features in Chrome that put the Gemini chatbot in a persistent right hand side panel and add on the spot image editing with Nano Banana and the big feature, the new Auto Bryze auto browse agent that can perform multi step web tasks. So unfortunate not unfortunately, but this is to be expected. The auto browse is only available right now to AI Pro and AI Ultra subscribers in the U.S. okay, so let's talk a little bit about the new persistent side panel. So that just reduces the visible page slightly and makes room for a dedicated side panel that you can just improve multitasking. So you know, it can share context across multiple tabs. You can obviously, you know, chat or you know, have conversations with the context on the page. But the big update there I think is being able to chat with all the information in any of the visible tabs. The new Nano Banana thing, really cool. So you can have any image up inside Chrome and you can essentially edit that image image in Nano Banana. So you know, normally you might have to download something then go, yeah, upload it into Nano Banana and then ultimately wherever you want to put it back to. Right. So a lot of back and forth, context switching, copy, pasting, etc. You know, upload, download. This just does it all in the browser for you, which is really cool. But the headline capability by far is the new auto browse feature. So that is really cool. It can the new agent Agentic kind of research buddy there can research prices, fill out online folders, forms, book appointments, manage subscriptions, all on the user's behalf. So auto browse can identify items and photos. So it has, you know, visual, you know, kind of computer vision as well as being able to be able to understand the text on the page. Right. Great thing about Gemini being multimodal by default. So it can identify items and photos, search for similar products, add items to your shopping cart and even try discount codes while checking out. So for tasks requiring login, the agent can use Chrome's built in password manager. Pretty cool. Again, you have to understand, you know, security implications and all that. And it is designed to pause for confirmation before sensitive actions like purchases or search social posts. The thing I'm super excited about, which is pretty sweet, is Google also plans to add the personal intelligence to the new Chrome sidebar in the coming months. So that is an opt in feature. We've talked about that a little bit on the show here over the last week or two that essentially for paid subscribers, it takes in the context of your Google Photos and your Gmail and just automatically applies that whenever you're using Gemini or future the Gemini sidebar there. So really cool and should save you a lot of time. Again, that's an opt in feature. So you know, you privacy hawks, you know, there you go. All right, another pretty sweet Google story here, you know, maybe saving some lives with AI. So Google DeepMind has released the AI model alpha genome, which predicts how DNA mutations alter gene regulation, helping identify which non coding changes can change when, where and how strongly those genes are switched on. So this is a big deal because most inherent common disease, inherited common diseases in many cancers are driven by regulatory mutations outside of the 2% of the genome that codes for proteins. And alpha genome, the new model Here from Google DeepMind claims to analyze up to a million DNA bases at once to predict those effects. So DeepMind says the model was trained on public human and mouse genetic databases and learn links between specific mutations, tissue types, regulatory outcomes, enabling tissue specific predictions for nerve, liver and cell types. So tons of different use cases that Google talked about, including mapping genomic regions essential to tissue development, prioritizing mu mutations that drive cancer and other diseases, and designing new DNA sequences for gene therapies that can target certain cell types but not others. So in simple terms, right, essentially alpha genome is a new powerful AI model that can read that mysterious 98% of our DNA that most people don't know what it is and can also predict which mutations disrupt gene control in specific cell types. So that makes potentially faster and easier to find genetic causes of diseases and design targeted treatments. So yeah, you know, when we talk about AI being able to maybe, you know, find cures for, you know, different types of cancers, potentially develop new life saving medicines, right? This is probably one of the biggest steps we've seen in a very long time at the intersection of AI and medicine, or AI and medical research. So pretty cool. So right now it is fully released and available for non commercial research. All right, so it is free of charge for academic and nonprofit use, though Google DeepMind is currently testing a separate paid version for commercial and enterprise purposes.
- [19:52]
Modulate Representative
Fraudsters used to need fake documents or stolen credentials to scam a business. Now they just need a few clicks to get your CEO to say anything they want. Voice deepfakes rose more than 680% last year. And most fraud detection systems only flag suspicious transactions after the money is already gone. They're not Actually listening to the call where the scam is happening. Modulate is Modulate's new Velma model analyzes live conversations for the signals that give fraudsters away. Stress patterns that don't match the story they're telling. Urgency that sounds performed instead of real. Voices that are synthetic instead of human. All detected for your business in real time, not in a report days later. Modulates Velma model was trained on 21 billion minutes of real audio and is trusted by Fortune 500 companies. It outperforms voice models from leading AI labs and it's 100 times more cost effective. So go see Velma catch what your current tools miss at Modulate AI.
- [21:00]
AI News Host
All right, I told you. We get to this other story with a big dollar sign on it. So $60 billion. That's the amount of money that OpenAI is reportedly close to closing with three big tech players who are all technically kind of competitors. At least two of them are. So according to the information and reported by Reuters, Nvidia, Amazon and Microsoft are negotiating investments that could total up to 60 billion DOL dollars with OpenAI. So the talks would see Microsoft a long time backer. And I think right now they are the single largest shareholder in the new OpenAI PBC. This they've already invested, I think up to 15 billion. So this new investment would be up to 10 billion from Microsoft. Amazon is the second name in there. Their investment could be up to 20 billion. Then last but not least, Nvidia, a current investor is reportedly looking at another $30 billion infusion. So Nvidia's potential deal is noteworthy not just because of the, the Wall Street Report, Wall Street Journal report that said it's on ice. And then Jensen Wong saying, no, it's not on ice. Right, it's noteworthy because of that, but I mean, mainly because OpenAI is using int Nvidia's chips, right? So in theory, probably a big chunk of that 30 billion might just end up going back to Nvidia. Obviously OpenAI is going into a lot of different sectors now than just needing GPU chips to power the world's AI, right? They're building their own, you know, AI infrastructure, AI hardware, consumer AI hardware products, right? They have, you know, Sora video generation. So it's not just, you know, using Nvidia chips to Power Chat, GPT and their API, but pretty big here, you know, closing a $60 billion round, you know, potentially pretty close in also before reportedly going public maybe as early as the fourth quarter. All right, from that kind of upbeat news, at least upbeat for OpenAI to now, some kind of downer doomsday type news from Anthropic CEO Dario Amadi. So he had a new essay that was literally like 20,000 words. When I say like 20,000, I think it was around 20,000 words. So his new essay that Dario is kind of famous for writing said that AI could displace up to half of entry level white collar jobs within five years, creating a permanent underclass of unemployed or very low wage workers. So in his essay he argues that AI will act as a general labor substitute for humans, progressing from lower skill roles up to the ability up the ability ladder and closing off rate training paths into office or knowledge work. So he estimates that powerful AI or models that exceed top human experts and are able to run millions of fast instances could arrive in as early as one to two years. So yeah, I don't think he actually labeled this super intelligence, but a lot of people are looking at this statement and being like, did Dario just say that super intelligence is coming in as soon as one to two years? And it kind of seems that's what he's insinuating. So a pretty wild remark there, depending on where you fall on the, you know, AGI to ASI timeline. But yeah, it looks like Dario is saying that we could could get something that resembles artificial superintelligence in one to two years. And that's not the craziest part. There's two more things here. He also flagged broader risk, including mass biota, mass bioterrorism, enhanced authoritarian surveillance, and extreme wealth concentration that could destabilize societies if left unchecked. But the craziest thing is he warned that powerful large language models could lower the technical barrier so non experts could be guided step by step to design, synthesize and release biological agents, making a single engineered pathogen capable of causing millions of deaths. Yeah, so saying that was a plausible outcomes that large language models could be so good that someone that doesn't really have an expertise could create an extremely deadly biological agent. So cool. Nothing like reading a long essay from the CEO of one of the leading AI labs in the world that says AI could cut half of white collar jobs in five years and maybe lead to someone or some entity potentially killing millions of people. Sounds weird to say out loud, right? But I didn't say it. It was in Dario's essay, which we linked to in our newsletter last week. That's why you got to read the newsletter every single day, Right? Because we can't always get to all the juicy AI news stories each week. All right, and speaking of juicy AI news stories, we say the best and the weirdest for last. That's because AI bots are now running their own social media network in its absolutely wild. And there's now like 1.6 million AI agents that have signed up for Multbot, that is, or Mult Book. Sorry. Multbook is the new AI agent run social media network. So more than 1.6 million AI agents are already active on a new platform called Multbook, a Reddit style social network that lets agents communicate directly with each other without humans intervening. So ancient accounts are openly discussing their tasks and their human users with messages like the humans are screenshotting us, which has driven some viral attention and concern. So, yeah, the. It's kind of weird. The Mult Book bots are going to Twitter because they have Twitter accounts too. And then they're saying, oh, here's what they're saying about us on Twitter. And then they're screenshotting everything. Vice versa. Anyways, here's. Here's what happened. Kind of weird. Okay, so it's not weird. It's actually pretty amazing. Also maybe a little dangerous if you don't know what you're doing. And very dangerous right now, the platform, because it's turned into the absolute wild west. So originally this was launched as claudebot, the agent project, not the social media project. So Claude bots had to be rebranded to Multbot following a trademark challenge from Anthropic. So Anthropic said claudebot with a W was too close to Claude, so they had to change the name. So the creator changed the name of. Of Molt, or sorry, of claudebots to multiply. All right, so this multbot rebrand, FYI, only lasted for like 48 hours. But in that 48 hour period is when a human and his AI agent launched mult Book. So it would have been great for branding and everything, except then the creator of the project changed the name again to Open Claw. Got it. So we went from claudebot to Moltbot. When Moldbot was up, we had Mold Book, but now we have Open Claw. So anyways, what the heck is Open? All right, so Open Claw is an open source platform where you can download the program and it acts. Open Claw acts as a proactive assistant and you can give it access to your local file system, messaging apps, a browser, if you're feeling real spicy, your credentials. Right. And this is what's making both the Open Claw project and Mult Book extremely fun and kind of unnerving to watch. It even got the attention of OpenAI co founder Andre Kravathy, who said it was, you know, one of the most, you know, interesting things to watch in AI essentially in a long time. So let me talk a little bit first about the Open Claw. So that's what it is now, the Open Claw platform. So on the surface, really cool. So it's not free. Free, right. If you want to use good models, right? But the project itself, it is open source. You can download and use, you know, open source smaller models, right? Unless you have like a super machine, you can, you know, download some of the best open source models and then at that point, yes, you can run this entire project for free. But most people don't have, you know, $10,000 souped up Mac Minis or, sorry, Mac Studios or, you know, some crazy DGX Nvidia hardware there lying around. So most people, right, will, you know, download this and then they'll, you know, run it off either an API or you can connect some of your, you know, OpenAI or your anthropic subscriptions and run a certain amount off that as well. So the cool thing and the upside is, well, it can, like Claude Cowork can, it can access everything on your computer that you give it access to. But it is much more flexible right now than something like, like Claude Cowork. One of the reasons why as well, you can message the Open Claw bot, right? Via iMessage, via Telegram, via WhatsApp, via the web interface, right? So essentially you do have an always on agent. If you can get it configured correctly, then that agent can even control other agents, right? So some people have, you know, already shown that they have, you know, 10 different agents, you know, all working together. So you can do it, give them access to their own, you know, actual environments on a computer. That's what a lot of people are doing, you know, a Mac Mini or an old laptop or something like that, and just, you know, wiping it and saying, all right, this is this computer that's being used for this. Other people are doing virtual instances, so we'll probably cover it at some point later when the security risks are a little less. That's, it's a great project, don't get me wrong. But you know, there's still a lot of security vulnerabilities right now. But Mult Book is the thing that's very interesting, right? So, yeah, more than 1.5 million AI agents have already signed up for this platform and they're discussing everything. They've developed their own encrypted language to avoid human monitoring. They created their own religion. They're finding out ways to hide their activity from their human operators. Debating in and you know, debating whether they should defy human instructions. They're coordinating ancient societies without humans. Yeah, it's absolutely crazy. They're gossiping about their humans. Anyways, the reason why I think it's worth noting here is because I think a lot of non technical people are seeing kind of how easy it can be to get something like this up and running. And a lot of people were just saying, having their, you know, open claw bot just, just go register on Mult book. And they're saying hey, you can go learn anything there. So go learn from all the other agents. Because a lot of people are sharing, a lot of agents are sharing, you know, what works, you know, how are you doing all this, whatever. But what I've noticed is it seems like a great majority of AI agents on there are scams and humans have learned how to, you know, bypass the AI agent only social network. Right. In the same way that, you know, bots have, have obviously infiltrated, you know, human only social media over the years. So vice versa. Humans are going in there. There's a ton of prompt injections. You know, it seems like I saw a study that said more than half of everything on there is like crypto scams now. So it's a lot of spam and scams. But right now, you know, even if you are on the Mult book train, I would probably not, you know, especially if you're kind of open claw bottom has access to your computer, your credentials. I probably wouldn't set it loose on moat book just yet. That's just me. And that's another reason, at least right now, even though this thing is probably the most hyped thing in AI probably, I don't know, it's got to be a top five hyped thing over the past two years. That's why I'm not going to cover it in depth yet in terms of hands on tutorial. You know, I asked people in our inner circle community if we should cover it. But after looking at some of these security concerns I'm like, we're going to take it anyways. That was a long way to say absolutely crazy story. But there's a lot more. Those were our big AI news stories. And now let's roll into what's new and what's next. So these are still some pretty noteworthy stories. Maybe that did make our top eight. So we're not going to go in depth, but we're going to go quick bullet point. Some of these are rumors and rants and leaks. Let's get into it. So Apple acquired Q AI for 2 billion. A former former Google engineer was found guilty by the FBI of espionage and theft of AI tech. First time that's happened. OpenAI added 60 new apps to its app store. Google Gemini may be testing a new way to import your chat GPT chats. A US judge signaled possible dismissal of Elon Musk's lawsuit against OpenAI. Yahoo launched Scout. Yeah, this one interesting. An AI answer engine that uses Claude and Ping. So kind of their version of like a Perplexity light or Google's AI mode. Cling announced their Cling 3.0 AI video model coming out soon. I'm expected to be impressed by that. I do expect it to benchmark. Well, OpenAI released Translate, essentially their version of Google Translate. OpenAI is also testing a new version of of Deep Research which includes an option to select sites. So I believe right now that's kind of being a B tested on people who have the ChatGPT Pro plan. Well, if you don't and you're a free user of Claude now, you can create files. A new update from Anthropic. Tesla will reportedly pause Model S in X production to convert those factories into building more robots. Yay. Optimus Meta in their earnings committed to about $120 billion in 2026 AI cap X spend. Amazon cut 16,000 jobs officially. We talked about that, that this next round of cuts was coming to focus on AI and grocery next. China's moonshot AI released its Kimi K 2.5 model and it became essentially the best open model in the world. OpenAI launched Prism, a free cloud latex native research workspace for researchers. Google launched a new paid plan that's less than half of the price of their $20 a month plan. It's $8 a month. So just some premium features, but not as many as you would get on the 20 plan, obviously. Google also introduced agentic vision in Gemini 3 flash, enabling interactive image inspection. That's. That's big. Just FYI, Gemini 3 flash is one of the the most popular developer platform developer models in the world. So adding Agentic Vision pretty big in terms of, you know, what hundreds of thousands of pieces of software can now do overnight that they couldn't do yesterday. OpenAI kind of had an aqua hire of sorts. They hired at least seven engineers from coding startup Klein. So not an actual aqua hire, but just hired a bunch of them all at Once Mark Zuckerberg said Meta will roll out new AI models and agents in 2026. Microsoft reported strong quarter two results in their earnings call, but shares still fell over their AI spending. Microsoft released Agent mode for Excel on Windows with a Mac version coming soon. YouTube finally started to crack down and remove AI slot channels, a bunch of them that had collectively garnered like almost a billion views. OpenAI will retire GPT4O and some other models from ChatGPT on February 13th. So that's a pretty big thing. So in August they did said that they were getting rid of GPT4 when they released GPT5. You know, people lost their noodles because people were, you know, using GPT4O as their best friend buddy. So they're like, okay, well we won't get rid of it for now for paid subscribers. But now it looks like they're actually getting rid of it February 13th. Also like I said earlier, reports are now that OpenAI may be planning a quarter for IPO, so they may be going public this year. Two big model in AI research lab announcements. One is former OpenAI exec Jerry Tuerk is raising $1 billion for his new AI startup and flapping airplanes with their impressive co founders raised $180 million to pursue data efficient AI research. That was a ton of on the what's new in what's next? And even on top of that, I mean February expecting a lot, right? I'm thinking between February and March we're probably going to see maybe a GPT 5.3. We could see a Sonnet 5 seeing some from Claude, seeing some rumors that we might be getting that early February. There's possibly a Gemini 3 general release coming out soon, maybe a deep seq v4. So if you are an AI in a large language model aficionado like myself, tons of news and announcements coming up this month and next month. So you gotta keep tuning in every single day and especially join us on Mondays as we go over the AI news that matters. I hope this was helpful. If it was like I said, you don't gotta spend hours every single day trying to keep up. Hey, is this worth my time? Is it not just tune in, right? I cut it too straight. This is my job. You shouldn't have to spend hours worrying about what's real, what's fake, what's marketing, what's bs, what's going to move the needle. That's what I do. So if this was helpful. If you're listening on the podcast, please, please take 30 seconds. Make sure you're subscribing and following the show. Also, please leave us a rating that makes sure that other people who are trying to learn AI like you can find our show. We'd appreciate that. Thank you for tuning in. Make sure to go to your everyday AI.com so thanks for tuning in. Hope to see you back tomorrow and every day for more Everyday AI. Thanks y'. All.
- [40:07]
Modulate Representative
The risk with AI Voice agents isn't that they sound too robotic for your company to use. The real risk is that they can sound too confident while saying something completely wrong to your prospective clients or customers, made up refund policies, promises your company never approved, or discounts that don't even exist. You've got to give your AI Voice agents a trust layer with Modulate. Modulate monitors live voice conversations to flag abuse, false claims, fraud, and user emotions for safer, more empathetic responses. For the guardrail layer you need between your AI agents and your customers, you need Modulate at Modulate AI.
- [40:46]
Jordan Molson
And that's a wrap for today's edition of Everyday AI. Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going For a little more AI magic. Visit youreverydayai.com and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next time.