EP 630: OpenAI brings Apps and Agents to ChatGPT, Google drops Gemini Enterprise, is the AI bubble here and more

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.

OpenAI is bringing apps and agents to ChatGPT without the need to code. Google Gemini kind of had a shocking surprise with their announcement of Google Gemini for Enterprise, which I think is a huge step forward. And is there an AI bubble and is it going to pop? Yeah, there's a lot of new updates happening and hot take topics circling the AI news world this week. If you missed any of it, Don't worry, we're gonna cut it to you straight this week with the AI news that matters. What's going on, y'all? My name's Jordan Wilson and welcome to Everyday AI. This is your daily livestream podcast and free daily newsletter helping everyday business leaders like you and me not just keep up with what's happening in the world of AI, but how we can make sense of it and actually use all this decision to make the right choice to grow our companies and our careers. If that's what you're trying to do, Awesome. Maybe it's your first time listening. Welcome. It starts here with the unedited, unscripted livestream podcast. But if you want to take it to the next level, that's what our website is for, youreverydayai.com. Go sign up for the free daily newsletter. Each and every day, we both recap the highlights from that day's livestream podcast, as well as bring you all of that day's AI news. And hey, also on our website, there's like 630 now podcast you can go watch, view, listen to, and learn from the world's leading experts, all for free. Also in our newsletter today and in the Spotify poll. So Spotify users, go scroll down in your show notes and check out the poll for today. I work for you. On Tuesdays, we do Hot Take Tuesdays. Mondays, we bring you the AI news that matters. So go vote for our Hot Take Tuesday episode for tomorrow's show. On Wednesdays, we do AI at Work on Wednesdays, Thursday and Friday, for the most part, bringing you interviews with those leading experts. But let's get into it. There was a ton happening in the world of AI news this week. Good morning to the livestream joining us. Brian from Minnesota, Larry joining us from NYC. I was actually in NYC last week. Secret project. I'll let you guys know what was going on. Everyone else joining on the YouTube machine. Jay joining from West Virginia, Big Bogey Face, everyone else. Dino joining from Italy. Thanks for tuning in. All right. Uh, first let's start with this one. Apps. All right. Uh, OpenAI is bringing apps. So in the same way, think of your phone. Uh, I want you to think of your phone and ChatGPT the same for a minute. Do you remember your phone before there were apps? Uh, smartphones at least? Yeah, I kind of do. And they weren't super, uh, helpful in terms of at least what you could do for your day-to-day personal and business lives. Well, now, and pretty soon, we're going to be living in a world where there's ChatGPT apps. So let's go over this new piece of AI news. So OpenAI, at their Dev Day conference, launched their new apps integration feature, letting users interact with apps such as Spotify, Zillow, Canva, Figma, and more in real time. So very limited rollout of what apps are available, but OpenAI is opening this up to third-party developers, and they said there's going to be a lot more apps rolling out in the future. So essentially, you can use ChatGPT to interact with these apps in real time. So the new capability right now is available to all logged-in ChatGPT users outside of the EU, including both free and paid accounts. So I've looked in, I think right now I'm using 8 or 9 different accounts. It's rolled out to about 5 or 6 of them right now. So if you haven't seen it yet, it's adding the same way that you would previously add connectors. So now it might be called sources depending on what tier plan you're on, but essentially you can add these apps. So let me just tell you real quick what this means. So let's talk about Canva because I think that's an interesting one and this kind of explains what's new. So previously Canva had had a GPT Canva had an official connector, but essentially these couldn't iterate and interact with the actual app in the, in the interface of ChatGPT. So now you can, you can actually not just create designs in Canva in the body of ChatGPT, but the real power here is it can use the context and understand the context of your current conversation to build, you know, maybe a pitch deck. That was one of the demos that OpenAI did. So you can work with other modes of ChatGPT, then call on, in this instance, the Canva app and build anything you want and iterate on it as well, just with natural language. So I do think the future, this is going to be huge, a huge step in ChatGPT going from an AI chatbot to what I've been saying for years. Large language models are trying to be your AI operating system. So think of, you know, maybe in 6 months or a year, when most of the websites that you use on a daily basis, because this is opened up to third-party developers, when most of those websites, you can interact with them, interact with the graphical interface, you know, I think it's really gonna change how we work, and that's not an exaggeration. And I do feel that the other big players, including Microsoft, Google, and Anthropic, are going to have to follow suit because these apps and the limited ones that I use change how you work. They really do. You know, so another easy example, right? The, the Zillow app. You can interact with the Zillow app in natural language. Say, hey, I, you know, I live in Chicago. I can say, hey, I want to find a new house. This is my criteria. And maybe there's filters for that in the app. Maybe there's not. And that's the other big thing. It's really being able to take this unstructured approach to working with apps through natural language, but combining that with your context of the conversation and being able to work with it. So pretty big deal from ChatGPT with that announcement. Speaking of big deals, this one maybe didn't get the headlines that a lot of the other stories got, but this is huge. So Google, at its Gemini at Work 2025 event, Google announced Gemini Enterprise and Gemini Business, marking its biggest push yet into enterprise workplaces. Replace AI and setting up a direct challenge to Microsoft Copilot and ChatGPT Enterprise. So I did cover this more in depth on episode 629 on Friday. So make sure you go back and listen to that. I got early access to Google Gemini Business, so was able to walk you through it. So make sure you go listen to that episode. But this is pretty big. So the new Google Gemini Enterprise platform is priced at $30 per user per month, which is pretty similar to Microsoft Copilot 365, but actually half of the price of ChatGPT Enterprise. So going after a head-to-head kind of showdown with Microsoft 365 Copilot, maybe those not satisfied with the platform, or maybe those still choosing an AI operating system for their enterprise, or maybe cutting into those that maybe don't want to spend that much on ChatGPT Enterprise. So Gemini Enterprise offers unified access to Google's top AI models, obviously including Gemini 2.5 Pro, which has ranked as the world's most powerful model in many third-party benchmarking services. The platform also consolidates a no-code drag-and-drop agent builder, enterprise search, and governance into a single interface. So Google is essentially trying to go with this new kind of AI one front door approach to bring people in because right now Gemini, it is kind of all over the place in a good and a bad way. I like to say there's many side doors and back doors to use Google Gemini. So what Google is trying to do is they're trying to kind of rebuild and say, hey, Google for Gemini for Business, Gemini for Enterprise. This is the front door because it has been confusing because even right now, as an example, my same workplace account. So, you know, if you use a Google address for your company, you can technically still have a personal Gemini account with that email and a business or an enterprise account. And actually, the two versions of Gemini work very much differently, which is also confusing. But the big feature in the Gemini for Enterprise and Gemini for Business, which is why you've got to pay attention, y'all. I don't want to repeat this 20 times, but let me say this. Grounding, grounding, grounding. It's huge. In the same way that myself and many people, millions of people love NotebookLM for its ability to ground its responses in your data only. So Google Gemini for Enterprise and Business has a similar grounding feature where the AI bases answers strictly on company data, and it reduces hallucinations and increases Trust, making Gemini Enterprise more reliable for those sensitive business tasks. And similarly to just about any AI, you know, large language model or AI chatbot, they do work with some third-party connectors. So on the Enterprise plan, Salesforce is a big one that I think a lot of people are going to like and use. You have Jira, Confluence, and also Microsoft 365 products. Big news right there. Livestream audience, let me know, if you want me in the future to do a comparison between Gemini kind of enterprise or business versus the personal version of Gemini, because there are technically big pros and cons to each. I do hope that Google essentially rolls out some new or some of their features that are great inside their personal product inside the business tier, because I think there's at least some of my day-to-day drivers of what I use Google Gemini for are not currently available in the business enterprise plan. Yet the grounding cannot be overlooked for teams that are serious about AI, but also serious about using your company's data. All right, uh, Annie here, uh, says if you use Teams, Copilot is nice. If you're using, uh, Google Workspace, Gemini works with it nicely. Yeah, it, it is ultimately about like what your organization use, but I was kind of surprised that Google Gemini Enterprise and Google Gemini Business rolled out with native Microsoft 365 integration, not across every single app, but across some of the main ones. So you, you have to, you have to be able to call that out. Right. So Jay here saying, I would love to hear more differences between the Gemini Personal and Enterprise. Same with Microsoft, honestly. Yeah, it's hard for me to do Microsoft because I'm not part of a Microsoft organization. It's a little bit easier for me to do the Gemini side, but I am going to have some some guests probably in October and November from Microsoft on to talk about some of their new agent capabilities. All right, our next piece of AI news, $20 billion of fundraising, but don't worry, it's not actually fundraising. So Elon Musk's AI startup XAI is reportedly doubling its latest funding round to $20 billion according to Bloomberg. So XAI's latest funding round now totals $20 billion, combining 7.5 billion in equity and 12.5 billion in debt. So the deal uses a creative structure, and part of that is with Nvidia. And we're gonna talk about this a little bit more when we talk about the AI bubble and circular funding. But, uh, the deal uses a creative structure. It's a special purpose vehicle, uh, where Nvidia will invest up to $2 billion, and essentially XAI will buy, uh, Nvidia AI chips and then lease them, uh, back after 5 years, meaning XAI does not directly take on the debt. Uh, so this approach gives financiers collateral in the form of hardware while allowing XAI to scale up its computing power without traditional corporate debt on its books. So NVIDIA, the world's leading AI chipmaker and the company driving the generative AI revolution, is reportedly investing around $2 billion in the equity portion as well, joining Apollo Global Management, uh, Diameter Capital Partners and Valor Capital. Uh, so the capital will fund, uh, XAI's next major data center called Colossus 2, located near Memphis, which Musk says is key to developing next-generation AI models. So the Wall Street Journal calls this the quote-unquote most expensive corporate battle of the 21st century as companies pour tens of billion dollars into AI, uh, infrastructure. Uh, Bloomberg reports that XAI is burning close to a billion dollars a month, although Elon Musk has publicly denied this and insists that X, the company, is not even raising capital right now. So, yeah, it's the $20 billion fundraise that reportedly just didn't happen. All right. More big AI news, and Microsoft is planning to kind of blow up and rebuild one of its major products made for developers. So according to reports from Business Insider, Microsoft is moving quickly to revamp its GitHub platform in response to rising competition from the new AI-powered coding tools. So Microsoft is reportedly overhauling GitHub to make its AI coding tools available wherever developers work, not just inside one app. So the push comes as GitHub faces competition from fast-growing AI coding tools like Cursor, Anthropic's Claude Code, OpenAI's Codex, Lovable, bold, there's a ton. So Microsoft CEO Satya Nadella told executives that AI is erasing boundaries between apps, documents, and websites, and Microsoft wants GitHub to reflect that shift. So Microsoft is investing in upgrades to the core GitHub features, including GitHub Actions for automation, analytics tools to track code performance, and enhanced security for code storage. Microsoft is also working to ensure GitHub complies with local data storage rules, aiming to expand its reach. So Microsoft GitHub Copilot remains popular, but recent data from Barclays shows it has lost market share to cursors and others in key developer segments. Microsoft is also encouraging teams to use AI more internally, even considering adding AI usage as a metric in employee reviews. Interesting. That sounds like a Hot Take Tuesday show waiting to happen. Y'all, here's Here's my talk on this. I do think some of the AI vibe coding tools are starting to lose a little bit of traction, including a lot of the recent reports are showing even Claude and Claude Code is losing a little bit of traction due to some model issues that many have reported doing, you know, also stemming from developers just using it a little less. Groq Code has become very popular. Gemini's latest updates to their 2.5 models very popular, as well as more and more coders are flocking to OpenAI's Codex as they release a specialized version of GPT-5 specifically for Codex. So it's interesting here to see now Microsoft GitHub acknowledge this and say, yeah, we're not competing here. Let's blow this thing up a little bit and rebuild, especially considering Microsoft GitHub Copilot was one of the first kind of copilot coding tools. So they were one of the first to the market, but losing share. So the space is going to get wildly more competitive here in the coming weeks. All right. Our next piece of AI news. Yeah, there was 3 big announcements from OpenAI's Dev Day this week. We already talked about the first one, OpenAI bringing apps to ChatGPT. The next one is a lot of new updates to its API. So OpenAI has rolled out some major updates to its developer API. And here's the big ones, GPT-5 Pro now available in the API, Sora 2 for video and audio generation available as well, as well as a new lower-cost real-time voice model and image model. So the two headliners at least would be the new GPT-5 Pro model that is designed for domains where deeper reasoning and longer context windows are critical, including finance, law, and healthcare. I really only see GPT-5 Pro being used on the backend for highly valuable tasks, considering it is about, I think, 12 to 15 times more expensive than the GPT-5 version. So I don't see it being used widely, but for certain sectors where it matters, I think it will be used. Sora2, different story. This is already rolled out to dozens of other AI video startups. So Sora2 enables developers to integrate the OpenAI technology into their products to generate synchronized audio and video from just text prompts, bringing advanced multimedia capabilities directly into their API for the first time. So what the heck does this mean? Well, actually, I do. I should also note the GPT Realtime Mini model. All right. So this is the model that developers can use to create more dynamic and interactive platforms. So it dramatically reduces the cost of voice interactions and it makes real-time audio apps more accessible for startups and small teams. So developers can now build richer end-to-end experiences with text, voice, and video from a single provider. This right here is really OpenAI trying to compete a little bit more on the dev end with Google, which I think has been dominating the multi- modal experience because they've been the only real one with the access, right, with VO3, their video model, their Gemini Live integration via the API as well. So again, right now, in terms of for developers building, I think it's a two-team race unless you're strictly building for software engineering, coding development, you might be using OpenAI. Pretty, pretty big news here. And I think for most developers, again, it's a two-horse race. It is OpenAI and Google running away from everyone else. What does this mean for everyone else? Well, all of the apps that you use, right? All of the SaaS products you use, you may not know this, but there's a very high likelihood if there's any AI feature, they're using either Google or they're using OpenAI. So what this could mean is all of the apps that your company uses could now either get video generation, could get a much more powerful model in GPT-5 Pro building it all, or could get interactive features with the GPT Realtime Mini model. All right. Here's a stunning piece of news that you could call it tiny news, but I think it's gigantic. So because we're talking about tiny recursion models. It was a joke, wordplay. All right, so a new open-source AI model from Samsung's Montreal research lab is making waves for delivering big results with a fraction of the computing power and cost, according to a new research paper. So what is this model? It is a new brand or genre of models called a tiny recursion model, and it was developed at Samsung's Advanced Institute of Technology, and it contains Ready? 7 million parameters. Yes. Most of the models are tens of billions, hundreds of billions, or even 2 trillion billion parameters. So this new model from Samsung that achieved great results on certain benchmarks is like 10,000 times smaller. So it contains just 7 million parameters, but on some tests, it does rival or outperform models that are 10,000 times larger, including OpenAI's O3 Mini and Google Gemini's 2.5 Pro. So the new TRM model achieved a 45%, and this is the one that made waves, 45% on the challenging Arc AGI grid tasks, results that matched or beat much larger proprietary models in those domains. Yeah, not just OpenAI and Google's, like, every everyone's models. So the model's breakthroughs comes from a radically simplified architecture that has just two layers using recursive reasoning to refine answers step by step rather than relying on massive scale or complex hierarchies on the front end in the inference phase. So TRM's code is also now freely available on GitHub under the MIT license, which allows anyone from startups to researchers to use, modify, and deploy it for commercial or research purposes. Yeah. So what this means is in the future, I think we're going to see a ton of new edge, edge AI use cases. Y'all, I got the receipts. Since 2023, I've been talking about domain-specific small models are going to dominate the future. of enterprise, personal, and professional AI. A 7 million parameter, uh, model can run on literally anything, right? It's, it's so small, right? Probably on a, I, I should look this up, but probably on even smartphones from like 5 years ago, right? So a lot of these, uh, you know, more edge AI or on-device, uh, AI use cases, you'd need the most expensive phone that just came out, or, you know, you need a, a GPU that's enormous, right? Spending, you know, $2,000 on a GPU for your computer. This is small. It can run on anything, which really changes what's possible in the future. If this new technology does catch on, it's extremely, extremely exciting. All right. Let's keep it going. Choo choo. Nisiani, what's up? Joining from the Chi in the livestream. Good to see you. All right. Is there a bubble? Is it going to pop? What the heck is circular funding, right? Uh, yeah, another new wave of, uh, AI fears are circulating in Silicon Valley, in the media. So AI moves too fast to follow, but you're expected to keep up. Otherwise, your career or company might lag behind while AI-native competitors Leap ahead, but you don't have 10 hours a day to understand it all. That's what I do for you. But after 700+ episodes of Everyday AI, the most common questions I get is, where do I start? That's why we created the Start Here series, an ongoing podcast series of more than a dozen episodes you can listen to in order. It covers the AI basics for beginners and sharpens the skills of AI champions pushing their companies forward. In the ongoing series, we explain complex trends in simple language that you can turn into action. There's 3 ways to jump in. Number 1, go scroll back to the first one in episode 691. Number 2, tap the link in your show notes at any time for the Start Here series, or you can just go to starthereseries.com, which also gives you free access to our inner circle community where you can connect with other business leaders doing the same. The Start Here series will slow down the pace of AI so you can get ahead. Silicon Valley insiders are sounding the alarm in some new reports as concerns mount over a possible AI investment bubble, with major players and experts warning about overvaluation and risky financial practices. So here's the, the bullet point, and, and this might be one of the, uh, Hot Take Tuesday episodes. We'll see. So make sure you go vote in today's newsletter and on the poll on Spotify. So according to reports, AI-related companies are driving 80% of the gains of the American stock market this year, according to BBC News, sparking fears of unsustainable growth. Oh, weird. Another thing I told you all 18 months ago when I was the first one to do this research, right? At the time, I think it was about 70%. 6 of the companies, uh, had about, uh, 60 to 70% of the NASDAQ gains in 2024. So the writing's been on the wall for a while, but the rest of the world is finally starting to do the math. Uh, so global spending on AI is projected to hit $1.5 trillion before the end of this year, with giants like OpenAI, NVIDIA, Microsoft, Oracle, and AMD making massive deals. All right, so, uh, let's talk a little bit more. Is this a bubble? And some of the recent, uh, eyes and attention on renewed bubble talk has centered around OpenAI's recent deals as well as NVIDIA. But one of the biggest ones was OpenAI recently entered into a $100 billion— yeah, with a B. Uh, the, the TRM was a 7 million parameter model with an M. Here we're talking about a $100 billion agreement with a B between, uh, OpenAI and NVIDIA to build data centers using NVIDIA's advanced chips. And also OpenAI announced similar plans to purchase billions of dollars in equipment from AMD and also potentially becoming a major AMD shareholder. So, yeah, depending on kind of performance, OpenAI might in the future become a 10% equity shareholder in AMD, which is absolutely nuts to think about. So why is everyone worried about that? Well, experts and investors are worried that these complex financing arrangements, which many are calling circular finding or, uh, fund financing or, uh, vendor financing, where companies invest in their own customers, may just be artificially boosting demand and clouding true market value. So let's look at OpenAI. But OpenAI's revenue is growing rapidly, but the company has never posted a profit, raising questions about the long-term financial sustainability at one of the major companies in the middle of all of this, AI bubble talk. I will say this. OpenAI is the most valuable private company in the world.

Right.

We talked about this on last week's AI News That Matters show. The most recent valuation was at about $500 billion, but that was before a lot of all of this recent news. So I wouldn't be surprised, ready? I wouldn't be surprised if the next time that we get an OpenAI valuation, that they're valued at more than a trillion-dollar market cap, which would make them a top 10 biggest company in the US. So, you know, a lot of people are worried about, oh, you, you know, OpenAI's not turning profit. True. But the potential is literally unfathomable about how a company like OpenAI has gone from a startup that no one really knew in October 2022 to now one of the most valuable companies in the world. It is because the generative AI technology is beyond revolutionary. Uh, so we'll continue to keep covering and if this is, uh, an actual bubble. But also we saw the surge in AMD stock this week is seen as a sign by some that retail investors are just chasing AI hype reminiscent of a previous of previous market bubbles, as well as we talked about the massive infrastructure projects such as OpenAI's planned $500 billion, 10-gigawatt data center in Texas are underway, raising environmental and financial concerns over overbuilding. All right. Is there deadly AI, right? Should we be worried about AI killing us? Well, a very prominent name said Yeah, you kind of might have to take that into consideration. So our next piece of AI news, Google's former CEO has issued a stark warning about the risks of AI falling into the wrong hands. So Eric Schmidt, Google CEO from 2001 to 2011, told the Sifted Summit in London this week that AI models can be hacked to bypass built-in safety guardrails, potentially allowing them to be trained for harmful purposes, including learning how to kill. Yeah, nothing unsettling about seeing that headline come across as you sip your morning coffee when you read the Everyday AI newsletter late last week, right? So Schmidt stressed that all major tech companies currently prevent their AI models from answering dangerous questions, but he warned that these protections can be reverse engineered and removed by skilled hackers. So Schmidt's comments add to the growing concerns from tech leaders, including Elon Musk, who previously warned there is a non-zero chance AI could pose a Terminator-style threat to humanity. So Schmidt has repeatedly called AI an existential risk, saying it could harm or kill many, many, many, many people as its capabilities grow. Despite these warnings, though, Schmidt remains optimistic about the long-term benefits of AI, believing its abilities will eventually far exceed human capability. All right. Our last big piece of AI news for the week, OpenAI, their kind of third big announcement coming out of Dev Day. We split them up because I think all three were pretty big. Their apps, number one. Number two, the API announcements. And number three, their agent kit. Bringing drag-and-drop, no-code, low-code agent building to the masses. So OpenAI announced Agent Kit, a new suite of tools designed to make building AI agents easier and faster for developers, enterprises, and even non-technical people. So OpenAI unveiled Agent Kit during their Dev Day event in San Francisco, and it includes Agent Builder, their visual canvas for creating and versioning multi-agent workflows. Also, there's a connector registry, which centralizes management of connections across OpenAI products and supports integration with large third-party services like Dropbox, Google Drive, SharePoint, Microsoft Teams, and others. Also, they introduced ChatKit. So essentially, you can use AgentKit and the Agent Builder to build agents. And then ChatKit, which is now generally available, allows developers to easily embed chat-based agents into apps and websites. And that is the big thing, right? So for the most part, when you're using ChatGPT or if you're using their playground, if you're not a developer, it hasn't been always the easiest. If you want to bring agentic capabilities using OpenAI's developer platform, now it's pretty easy, right? In the demo, they did this in like 6 or 7 minutes. But I think the average person could, even if you're non-technical, you can get this in like an hour, right? Literally. Let me say that again. Even if you're a non-technical person, you can go use the Agent Builder and the Agent Kit, which includes Chat Kit. And Chat Kit is how you can deploy an agent on your website or your app. But let me just— let me say this straight up. Is the Agent Builder— is this actually an agent? Not really. I would say this is more AI-powered workflows. right? But we'll see where the capabilities go in the future. So Agent Kits also introduced Guardrails, an open source safety layer that protects against data leaks, jailbreaks, and unintended behaviors, addressing some enterprise security concerns. So like I said, competitive space right now. So we have Microsoft Copilot Studio, Google obviously with their new announcement of Gemini Business. And Gemini Enterprise rolled out their agent builder. Very limited. So Google's is very limited, but very impressive results. And I think it's almost the opposite, maybe, for OpenAI's. It was a little buggier when it came out, but much more robust. So I'm guessing once OpenAI irons out some of these bugs in their agent kit, it may be more capable than some of the other competitors. All right. That was a ton, but there's a ton more. All right. So those were our big stories, hopefully giving you a little depth and explanation and context, but so much happened this week in AI. I can't spend 60 seconds on every single story. So I'm going to give you all the what's new and what's next rundown. So some big stories, some new releases. Some rumors. Here we go, bullet point style, going through them. All right, so maybe if you're listening on the podcast, feel free to slow this part down. I'm going to talk fast. Here we go. So OpenAI has urged the EU to keep their big tech competitors in check. OpenAI could face a billion-dollar fine over pirated books, just like Anthropic faced. Speaking of Anthropic, they launched Claude Code with plugins. Figure AI has released their Figure 03 AI robot, which can finally do the dishes and laundry. Yay. Meta has hired Thinking Machines Lab co-founder Andrew Tolok reportedly for $1.5 billion. Yeah, billion. All right. Google released Gemini 2.5 computer use, which we covered in the newsletter last week. Open— sorry, IBM released their Vibe coding tool called Project Bob, Love that name, by the way. A new source code update shows hints of OpenAI's much-rumored browser codenamed Aura. So we'll probably see some fresh rumors hitting that this week. Copilot for Windows just got smarter with document creation and inbox connectors. Gartner came out with a new stat and study and said they expect 40% of enterprise software companies to be vibe coding by 2028. I think it'll be more than that and sooner. There was a Deloitte AI snafu with the Australian government where they essentially had to refund them a bunch of money because they turned over a hallucinated report and they used an old model. My gosh. Also, the AI bubble worries. We talked about that. IBM and Anthropic launched a partnership. AI startup Reflection AI raised $2 billion in new funding. GPT-5 Pro set a new Arc AGI record with state-of-the-art results. The browser company finally publicly launched their AI browser, Dia, and it's generally available to Mac users without a waitlist or invite code. Nvidia CEO Jensen Huang said in an interview this week that the US is not far ahead of China in AI, kind of sounding the alarm there. OpenAI released a new study that showed GPT-5 had a 30% less political bias versus their previous models. Google is rolling out Nano Banana image, image editing inside AI mode. Y'all, AI mode, once all these updates become generally available, I cannot underemphasize how big this is because Google CEO Sundar Pichai said that AI mode is going to become the default mode when you go to AI search. So all these capabilities, uh, eventually rolling out inside AI mode is huge. Uh, Google also announced that 50% of their code right now is being generated by AI. My gosh. ChatGPT rolled out new connectors, uh, for pro users, including Slack, Notion, and Linear. And there's a new US Senate bill, uh, here in, uh, the US to prioritize companies, uh, using AI chips and frowning on exporting them to rival countries. My gosh, that was a ton. What do you want to hear tomorrow? Hot Take Tuesday. So, uh, thank you for joining. On Mondays, we bring you the AI news that matters. Tuesday, Hot Take Tuesday. So make sure you check out today's newsletter and vote for what you want to hear tomorrow, or check out the poll in Spotify. New thing we're trying to do. So scroll down in the show notes on Spotify and vote on the poll. What do you want to hear? If this was helpful, please let me know. Repost this to your network. Let other people know. We spend so much time trying to make sure we give you an unbiased, no fluff, no BS news on, hey, here's What's happening and here's what matters. So if this was helpful, don't be greedy. Don't keep Everyday AI as your cheat code. Go share with someone. That's the only way we can keep doing this. So, uh, thank you for tuning in. If you haven't already, please go to youreverydayai.com. Sign up for the free daily newsletter. Thanks for tuning in. We'll see you back tomorrow and every day for more Everyday AI. Thanks, y'all.

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.