Ep 779: First big AI IPO launches, Anthropic gets called out, Google preps for big AI updates at I/O 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.

This past week in AI news may be the calm before the storm when it comes to a hot AI summer. That's for a few reasons. I mean, we finally have an AI company's IPO in the books with Cerebras, and it was a big one that will set the stage for Anthropic, SpaceX, and OpenAI. Speaking of SpaceX and OpenAI, the Musk versus OpenAI jury deliberations are starting today with a decision expected this month. But the even bigger AI news, at least when it comes to the tech that we all actually get to use, could be coming tomorrow with Google's I/O conference starting, which will undoubtedly set off the summer AI release frenzy. So new AI companies going public, old AI drama that's about to wrap, big AI releases hours away. A lot to cover this week in AI news. Let's get to it. Welcome to Everyday AI. If you're new here, my name's Jordan Wilson and we do this, well, every day. This is your daily livestream podcast and free daily newsletter helping everyday business leaders like you and me not just keep up with all of the AI developments. I help you focus on what matters, tell it to you straight so you can take that information to grow your company and your career. It starts here with the unedited, unscripted livestream podcast, but make sure to go to youreverydayai.com, sign up for our free daily newsletter. That's where we're gonna be recapping not just the highlights from today's show, but that's where you're gonna be able to sign up for our free daily newsletter that's gonna help you become the smartest person in AI in your company or department. All right, so let's get straight to it. First, sounds like a small update. This one, I'm telling you, is big. This is going to eventually, I think, take some of the, you know, OpenClaw and Hermes Agent momentum because Codex is becoming a beast. So here's what's new, and we did go over this in our Friday Features show. So that's episode 778. So just click the back button if you missed it. There's a lot of other AI news that was kind of more feature-related, uh, related that we aren't going to get to today. So make sure you go check that out. But we did go over this one on Friday. So here's what's new. OpenAI rolled out remote control access to Codex inside of the ChatGPT mobile app, allowing users to monitor and control Codex agents running on other computers directly from their phones, which marks a major shift in how people interact with long-running AI work. So access is available now. Now. It's working. I have it running. I'll go over the pros and cons here in a minute. So all you really need is a standard ChatGPT plan. You can even do this on the free version. It works best, obviously, if you have the paid version in terms of limits. So all you need is the normal ChatGPT app on iOS or Android, and then the Codex app running on any computer. Actually, right now, the remote control feature is not yet available for Windows. Uh, so according to OpenAI, more than 4 million people now use Codex weekly, and this update is designed to support longer, more autonomous workflows without requiring users to stay at their desks. So the mobile app now mirrors the live state of connected machines. Yes, that's with an S, machines, showing real-time screenshots, terminal outputs, file changes, test results, and approval requests, while all files, credentials, and permissions remain securely on the host computer. So users Users can approve actions, change directions, or add new instructions from anywhere, meaning Codex jobs can continue running while someone is commuting, traveling, or if you're just away from your primary workstation. So OpenAI says this is not just screen sharing or task control, but persistent access to full project context, active threads, plugins, and agent memory, which keeps context, which keeps complex work moving without restarting or losing progress. Here's my takeaway so far. I've been using this for a couple of days. Like Claude Dispatch, it is a little buggy. However, nothing like Claude Dispatch in terms of capabilities. So if you remember, Anthropic came out with a very similar feature called Claude Dispatch. The downside is Claude Dispatch only worked in Claude Cowork. It didn't work with your normal Claude Chat on desktop. It didn't work with Claude Code on desktop. And then the bad thing is, is you literally only had one running thread, which to me made no sense. So if you had all these different things running in Cowork, you couldn't really control them all. It was just one kind of threaded conversation, and mine broke literally all the time. Uh, so the remote control feature inside the ChatGPT mobile app with Codex is completely different. When I say it's a little buggy, I mean a little, right? Whereas, uh, Anthropic's, um, Dispatch literally I couldn't use it. I had uninstalled Claude on my machine, uninstalled the, uh, the, the iOS app. It just didn't work. So this is actually comparatively light years better. It is so good. Like, especially for me, one thing I really love with this is I have multiple, uh, you know, Macs that are running Codex on the same account. So the cool thing is, is you can connect your different, uh, computers as well. So I was firing off new threads. I was continuing, you know, old conversations with long-running agents. So things that are absolutely great. And again, I did a show on Codex last week, so don't think of the word code in Codex. This is literally, I think, a general knowledge worker's dream inside Codex. You know what? So one thing, you know, kind of some of the pros and the cons. Some syncing issues initially, and I think those got a little bit better. Also, for whatever reason, starting kind of new chats in Codex didn't always sync as well as picking up ones that had previously started from your actual machine. So I'm sure the OpenAI team is addressing those things. But this for me, and by far, is the first time that I feel we've had a general use case always-on autonomous agent that just works, right? Yes, there's OpenClaw, but OpenClaw requires like so much setup, right? A lot of like duct tape to get it to work, and something that was working, you know, a couple weeks ago might not be working as good now. Codex Remote Control in the ChatGPT app is wild, wild, right? And for me, it's, it's, it was kind of like nostalgic in a weird way because I hadn't really been using the ChatGPT app a ton because I've only been using Codex almost exclusively, right? So I went over that in the Codex show last week. So it was almost weird, like, going back into the ChatGPT app on my phone where, you know, maybe I'd go in there, you know, a couple times a week, whereas before, you know, I was spending a lot of time in there. So, all right, pretty big news regardless. All right. And our next piece of AI news, Cerebras. Yeah. So Cerebras Systems just delivered on one of the biggest tech IPOs in years with shares jumping 60%. 28% on the first day of trading Thursday after they went public and closed with a market cap of $95 billion, which is much higher than even some of the bullish analysts were talking about. So this is pretty noteworthy because it places a relatively young AI chipmaker just shy of the rare $100 billion first day valuation club, historically reached only by a handful of tech giants. So if you don't know Cerebras, well, they build kind of these wafer-scale AI chips and high-speed inference systems designed to train and run large AI models faster than traditional GPU-based infrastructure. So, uh, yeah, Cerebras talking about so close, right? $5 billion off in, in, uh, market cap. So they narrowly missed joining companies like Alibaba and, uh, Meta, which both crossed the $100 billion mark on their first trading day, with Alibaba closing its 2014 debut at more than $200 $131 billion. So for context, Facebook ended its first day as a public company in 2012. Yeah, that's crazy. It's like 15 years ago with a market value of roughly $104 billion despite reporting $3.7 billion in annual revenue at the time. So Cerebras momentum is tied in part to a multi-year contract with OpenAI valued at more than $20 billion announced in January and a partnership with AWS launched in March. So the strong debut is being closely watched as a potential signal that public markets are again open to massive tech offerings after a slower IPO period. Analysts say the Cerebras listing could help set the stage for potential blockbuster debuts from private heavyweights such as SpaceX, OpenAI, and Anthropic later this year. So, yeah, actually great news for those 3 companies due to the fact that Cerebras came in much higher than expected. They actually had to adjust their pre-market position because they were oversubscribed. There's just more and more people wanting in action, wanting action of their IPO in pretty, pretty high numbers, right? Because even if you go back and look at some of the, you know, kind of early rumors or early projections at where Cerebras was going to end at, you didn't have anyone talking about them potentially being a, you know, $100 billion market cap company on the first day. Yeah. So even the most bullish estimates were not really in that category. So this, if nothing else, is great news for the other 3 AI-related companies that are likely going public this year. The first being SpaceX, which we will probably see some, some official news on later this month. And then OpenAI and Anthropic expected either in Q3 or Q4 this year. So pretty monster debut for Cerebras, right? And if you don't follow the chip space, right, all I can say is go, go open Codex and go find, click the more models. And it's technically an older model, but go use GPT-5.3 Codex Spark. It is one of the wildest experiences doing computer use with GPT-5.3 Codex Spark, it's like, and go compare that to, you know, your Claude desktop computer use. And it's like watching someone that's never used a computer before in Claude desktop versus watching, you know, like a college graduate, uh, use it. It is that fast. It is that good. Uh, so big things, uh, ahead for Cerebras. All right. Uh, our next piece of AI news. Yes. Even before Google's I/O, they actually released a ton of at least hardware and some more information on their new Gemini intelligence. Uh, so Google announced a new category of laptops called Google Books at the Android Show, positioning them as a hybrid that combines Android and Chrome OS with Gemini AI at the center, signaling a major shift in how Google approaches personal computing. So according to Google, the core idea behind Google Books is Gemini intelligence, which is designed to offer proactive context-aware help directly on the screen, aiming to reduce the steps needed to complete everyday tasks. So the first Google Book laptops are expected to launch this fall and are expected to heavily compete with Apple's newly launched entry-level $600 MacBook Neo. So we don't have prices on the new Google Books yet, but they are expected to be in the MacBook Neo range. Uh, in one headline feature, which was really cool to watch, if you didn't check this out in our daily newsletter, uh, we'll link to it again, but the new Magic Pointer. Uh, so that allows users to select anything on their screen. Essentially, you kind of shake or hover your mouse over something, and then you can talk, uh, to the Google Books/Gemini intelligence, and it can understand what you're pointing at, and it understands the context of everything else on your screen, and then can take some actions on your behalf. So as an example, you can point at a date in an email and have it create a calendar event, or selecting two image images to visualize them together. There's another AI feature called Create My Widget, which lets users generate custom desktop widgets using information from the web and personal Google apps like Gmail and Calendar, including guided setup or text-free prompts. So whoo. Who's getting these new Google Books? Well, partners now on the hardware side include Acer, Asus, Dell, HP, and Lenovo. And Google says the devices will feature premium materials and a distinctive new glow bar, though its exact function has not yet been explained. So a little bit more on the underlying technology here, Google Intelligence. So yeah, this one's interesting timing from Google, not just the fact that, you know, they announced it the week before their I/O conference, but the fact that it's coming less than a month before Apple's WWDC conference. So yeah, remember the whole Apple intelligence thing that Apple couldn't deliver on, and then they had to pay a $250 million class action lawsuit for essentially lying to the public? Well, it's pretty, pretty interesting jab here from Google, uh, you know, essentially because it looks like Google's delivering Google intelligence over the course of like, hey, we're announcing it and it's coming out pretty much immediately. Uh, so it's highlighting the growing AI rivalry as both companies face pressure to deliver on long-promised personal AI features. So the Gemini intelligence, as we zoom out on what it is, what it does, and how it works, it bundles 5 AI-driven tools right now, including smarter web search in Chrome, AI autofill for forms, voice-to-text cleanup in Gboard, custom widget creation, and multi-step automation like booking classes or building shopping carts, all aimed at reducing everyday friction for users. So despite Google's AI head start, many features echo Apple intelligence capabilities first shown at WWDC 2024, but they're obviously still missing, raising questions about whether Gemini's promised summer rollout will meaningfully change how people work, communicate, or run small businesses. Uh, so. we'll obviously have more on Google's, AI plans with their upcoming I/O conference toward the end of today's news roundup. All right, next, a little bit of drama. All right, so if you weren't following this one, not a good look for Anthropic. Yikes. So Anthropic is facing this sharp criticism online, and that's putting it nicely, from developers specifically after announcing a change to how Cloud subscriptions handle programmatic usage, and that is set to take effect in less than a month, starting June 15th. So yeah, this was a huge mess. My gosh. So let's, let's unroll it if you missed it. And it is actually really important because I do think it is going to completely shift the landscape of how AI is used via the API. Anthropic obviously thinking this move is going to bring them in more revenue. I'm not sure. So we'll see. So here's what happened. So according to Anthropic, paid Claude users will no longer have programmatic usage counted against their normal subscription limits and instead will receive a separate monthly agent SDK credit ranging from $20 to $200. So essentially whatever plan that you're on, you know, whether the $20 plan, the $100 Max plan, $200 Max plan, you essentially get an agent credit. So this is different, whereas before, and with ChatGPT as an example, you can log into your ChatGPT account through a variety of third-party services and, you know, use your normal limits that way. The big difference here is, well, you're going to have to connect a card and you're going to run into some overage pretty soon. Also, the key change is that workflows using Claude P, or that's kind of the programmatic shortcut, the Claude agent SDK, GitHub Actions, or third-party Claude code tools will now draw from this smaller API metered credit pool rather than the broader subsidized subscription use usage that many developers and everyday users previously relied on. So that's the big shift here, right? So Anthropic was kind of saying, hey, nothing's going to change, right? You can still, you know, now you can use your, you know, your API in OpenClaw, you know, your Anthropic account. And everyone's like, no, not Really, right? Because before it was, well, truly subsidized. All right. Maybe this is a concept I'll tackle maybe in our Start Here series. Right. But for the most part, if you look as an example, what you can get out of a $200 Claude Max plan or a $200 ChatGPT, you know, Pro plan. And if you were to match that up with what that would cost if you were using it on the API, you know, depending on this, it depends on the service of the estimate, but normally you're at like like minimum 50x, right? You're going to get 50x more, um, on the, you know, quote-unquote subsidized plan than if you were just paying for the API usage. So Anthropic here— 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. This isn't my, you know, my wording. You know, this is everyone else online. They've been getting roasted, calling, pulling the rug, bait and switch, and just lying. Right. Developers are saying that Anthropic is lying to them and how they actually announced this in their announcement post and in the rollout. And well, I probably agree with that. So essentially, an Anthropic employee posted on X that users don't pay extra, quote unquote, under the new policy. And the post actually received a community note on Twitter, right, which is essentially a user-added fact check on X to design to add missing context or correct potentially misleading claims. So the community note explained that while subscription prices stay the same, programmatic usage that was previously covered by normal Claude subscriptions limits will now consume a separate $20 to $200 monthly credit at standard API rates. That's the big thing there, underscoring why many developers say how Anthropic has tried to spin this, and they're saying it's bad PR, a money grab, and a slap in the face to the devs that help Anthropic gain traffic over the years. So developers reacted strongly because Anthropic framed the change as a new benefit. Oh my gosh. When I read this, even before I looked at the comments and the reactions, I literally facepalmed. I'm like, I can't believe Anthropic actually tried to put this out as a new benefit when everyone knows this is a money grab, or at least, you know, they're trying to move away from subsidized pricing. So I guess you can't blame Anthropic for that, right? But if everyone else is still playing the subsidized game and Anthropic spins this, or they attempt to spin this as a new benefit, it's an absolutely terrible look when it comes to PR. Um, so, uh, even though, you know, many power users are going to effectively lose their significant included capacity or have to pay to maintain the same workflows. So critics argue the policy classifies usage by interface rather than intent, meaning that even human-directed workflows that run through the command line interface or the SDK are treated the same as fully automated agents. Yeah, that part to me, again, bonkers. So according to developer estimates shared on social media, some users could see a 10 to 40 times reduction in included programmatic usage compared to what they previously consumed under subscription limits. So the backlash was amplified by recent mixed signals from Anthropic, including, uh, what they said were higher usage limits announced on May 6th at their Dev Day, right? So they said, um, this one was, was funny. They said, uh, you know, higher, you know, 5-hour, uh, you know, limits, but your weekly limits stayed the same. So yeah, it's, it's, it's been a— let's just say May has not been a good month, uh, for Anthropic. And this one, yeah, they're gonna have to claw way out of it. Not to, uh, play on the pun here with the open claw, but you know, one of the reasons I think that, um, you know, Anthropic was able to gain its initial prominence, uh, in late 2023 and 2024 was being developer friendly, right? I actually had a podcast, I don't know, I'd say in the end of 2024, maybe it was early 2025, all the years blend together, but I essentially said that Anthropic had abandoned general business use cases because they were that, you know, heavy tilted toward developers. And I said, if Anthropic doesn't pivot in the next year, they're not going to exist. Obviously, they pivoted and they pivoted hard, you know, really going after verticalized industries such as legal, you know, such as finance, etc. Also now pivoting towards small businesses. with the recent offering there, but seemingly not just abandoning developers, but what looks like on paper just constant smacks in the face. So not sure how this one's gonna play out, especially as OpenAI is seemingly there, uh, you know, to pick up the pieces. Um, and obviously I think Google, uh, is gonna be continuing with what they may be announcing to scoop up some of, uh, Anthropic, what is they're gonna undoubtedly leave leave on the table. All right, our next piece of AI news. The United States and China held rare and formal talks in Beijing focused on setting guardrails for AI, reflecting growing concern in both countries about AI safety risks and global stability. So negotiations were led by US Treasury Secretary Scott Bessant, signaling that Washington views AI not just as a tech issue, but as a core economic and national security priority with long-term consequences for global markets and jobs. So a central focus of the talks was preventing non-state actors, including hackers and terrorist groups, from accessing or, uh, from accessing or abusing frontier AI models, as both sides discuss joint best practices to reduce the risk of catastrophic misuse. So President Trump said the US pushed for light-touch, high-level guardrails designed to reduce extreme risks without slowing innovation, reflecting concern that overregulation could hurt American companies and workers competing in the economy. Uh, Secretary Bassett said the US was only willing to engage in talks with China because it believed that it maintained a clear lead in advanced AI, allowing Washington to shape global standards in ways that favor US firms and values. But the elephant in the room was, well, semiconductor policy. And that has remained a point of friction as the summit produced no agreement on easing US export controls on advanced Nvidia AI chips, despite Nvidia CEO Jensen Huang attending as part of the US business delegation. The other couple other big things. Well, President Trump said China acknowledged the issue on the chips, but emphasized its desire to build domestic chip capabilities, underscoring the ongoing technology decoupling that affects global supply chains and hardware talks, uh, costs. And the talks took place against a pretty tense backdrop after a White House memo, uh, just about a month ago accused Chinese firms of running large-scale efforts to copy or distill American AI models, an allegation Beijing denied. So that is ultimately going to be, I think, the story that will decide whether these quote unquote AI talks between the US and China are just one time and more for show, to show that both sides are trying to do the right thing when it comes to ethical AI versus if they will actually be able to come to some sort of formal agreement. Right. So I think at this point, this is a good effort on both sides, but talk is cheap. This is actually a global concern, not just when it comes to the economic ramifications, but also the national security concerns for both countries. I mean, when you talk about military use, you can't overlook that as well, because I do think that most future wars will be fought cyber first. Um, so when it comes to distillation, model distillation, uh, from China, so, um, obviously the big AI labs in the US White House, um, have accused Chinese companies of distilling or essentially stealing, uh, all of this American innovation. So, um, I was maybe on a personal side, I was expecting, uh, the distillation to, um, be more of a focus because ultimately that supersedes everything else, right? Because you can talk about guardrails and chips and all this all you want, right? But if the US is going to continue to let China distill models, as the big tech companies have alleged and as the White House has confirmed, I don't see how these talks are fruitful at all. Right? And if I'm a CEO of a big tech company, I'm not looking at these talks even though they are rare, right? It was a good step for the US and China to meet, you know, definitely bitter rivals when it comes to AI. But yeah, if I'm an AI CEO looking at this and not seeing anything on the distillation front, I'm not too excited. I'm not too happy about that because these companies are losing billions of dollars a month, because of distillation. If you look, you know, even as an example at OpenRouter and just the, you know, the companies, the Chinese companies that have been accused of distillation. And if you look at just the trillions of tokens that they're, you know, charging just via OpenRouter and then the billions of dollars a month, what that's taking away from American companies. Uh, I'm not super happy, right, if I'm in their shoes. But I'm not. I'm just here talking about it. All right. Speaking of tensions running high. Yeah, here's one. OpenAI and Apple are apparently, well, sour apples. So according to Bloomberg reporter Mark Gurman and a new report as well in the New York Times, OpenAI is considering legal action against Apple. Raising questions about the future of their, uh, high-profile AI partnership. So the dispute centers on Apple's integration of ChatGPT in its Apple Intelligence features, which were announced in 2024. OpenAI reportedly believes that Apple failed to meet key terms of the agreement by making ChatGPT difficult for users to find and access within Apple Intelligence, and I absolutely agree with that, right? It was pretty, pretty hard. Anyways, uh, one of OpenAI's main expectations was that Apple's massive user base would drive new paid ChatGPT subscriptions, but internal results reportedly fell short of those expectations. So tensions increased after Apple confirmed in January that it has also struck a deal with Google to use Gemini to power parts of Apple intelligence, including the long-awaited AI overhaul of Siri. So Apple is now— or sorry, OpenAI is now reportedly weighing whether to serve Apple with a formal breach of contract notice, according to The New York Times, potentially opening the door to a lawsuit. So this news came shortly after Apple agreed to that $250 million settlement in a class action lawsuit over claims it failed to deliver promise AI features, though Apple did not admit wrongdoing. So if the partnership weakens or ends, it could affect how Apple users experience AI tools like Siri and how quickly advanced AI features roll out across Apple devices. So yeah, if you never saw or experienced this integration, it was supposed to be kind of seamless, right? So essentially, Apple kind of admitted that their Siri wasn't any good in that that, you know, if you asked it a basic question, their internal models would handle it. But if you asked it a more complex question, it was supposed to kind of seamlessly kick that query over to ChatGPT, which it didn't really do, right? I even, you know, manually enabled it and it barely worked, right? And so that's me, a paid ChatGPT user, trying to force, you know, Siri to use that, that ChatGPT integration versus using its own internal technology because the internal technology is, well, useless. So I can see this point from OpenAI's perspective, right? That because initial reporting said that there was no monetary, uh, tie to this, right? So it wasn't like OpenAI was getting, you know, a $2 billion or $5 billion a year deal, uh, for this, providing this service. Um, it was supposed to be that, well, Apple was supposed to kind of push people over, uh, to ChatGPT, and then OpenAI was, uh, expecting to convert a lot of those into paid users. So, uh, at least according to reports, it seems like Apple may have fallen short on that promise, just as they fell short on their ability to deliver Apple Intelligence that they actually marketed but didn't actually bring to market. All right. And our last big AI story, and this one is actually going to be playing out out, uh, tomorrow and Wednesday. But Google I/O conference is starting tomorrow, and it's shaping up to be a pretty big one. So multiple credible leaks suggest that Google is already testing a new version of their Gemini model with the 3.2 series, including a Flashlight Live variant showing up in the Google Cloud Console, indicating near-term production readiness rather than a research demo or something that might be available later. Also, there's a higher impact rumor that centers on Gemini Spark, described by Testing Catalog as a persistent 24/7 agent that could manage inboxes, website users that are— or sorry, websites that users are logged into, connected apps, and multi-step online tasks, which would mark a major shift from Gemini still being more, or I'd say less agentic than Claude and Copilot and ChatGPT. So moving away from just the chat-based assistant to the always-on automation. So yeah, Google inside Gemini, they have released some agent features, but they really haven't rolled out to the majority of their users. So we may see an announcement on Gemini Spark. Uh, so keep an eye out on that. We'll be covering it in our newsletter. Uh, so Google has officially also confirmed their Android XR glasses will be previewed at I/O following earlier teasers with expected partnerships involving companies like Samsung and XREAL, positioning smart glasses as a practical Gemini-powered interface rather than a concept device. Also, the Google Book, right? So we talked about that, is expected to get deeper platform and developer details at I/O. So not sure if we'll get pricing, but I'm guessing we should get some more hardware and software information on the new Gemini intelligence that powers Google Book. Also, Google says Chrome will become an AI agent surface with Gemini in Chrome handling tasks like browsing, summarizing, and image editing built on Gemini 3.1 and rolling out in the US starting in late June for eligible users. All right, that's not all of the leaks, uh, that we've heard because there's also the Gemini Omni. So yeah, we could get a refreshed, uh, video generation, uh, model. So, uh, reports have called this Gemini Omni and it's appeared in multiple leaks reported by outlets like Gadgets 360 and Testing Catalog, but Google has obviously not confirmed these features yet. So more of a multimodal version versus it just being, you know, v03 or sorry, v03.1. So we may see a newer version of Gemini that has more multimodal capabilities built into it all in one. So last year at I/O, Google rolled out Gemini 2.5 Pro and Flash, v03, Flow, as well as the Google AI Ultra plan, AI mode, and a lot more. So I would expect expect a ton of announcements from Google. Uh, we may see, uh, you know, Gemini 3.2 Pro, although we haven't really seen a lot of leaks on that. Uh, we have seen or heard, uh, through various sources that there may be a much more cost-efficient version of Gemini 3.2 Flash, uh, which would really cut into Anthropic's, uh, kind of revenue stream there. Uh, some of the reports are saying meaning it may be about 95% at Frontier level for about 20, about, I think, 10 to 20x more efficient. So if that is the case, I mean, that's going to take a huge chunk out of Anthropic's book. But, you know, Google is a big investor in Anthropic, so I guess in that case, if that is the case, they'll win either way. But Anthropic may not like that one. All right, that's not it. That's our big stories. But let's quickly go over the what's new and what's next. Uh, some of these are, uh, rumors. Some of these are smaller updates that have already dropped. Some other AI news stories that didn't make our, uh, big cut, but here we go. So, uh, OpenAI launched Daybreak for AI-assisted cybersecurity, kind of their version of Mythos from Anthropic. Uh, OpenAI also launched a connected personal finance, uh, platform inside of ChatGPT right now, only for pro users. Uh, the Musk-Altman trial testimony has concluded. Included, and a federal jury will begin deliberations today. Anthropic and PwC expanded their enterprise AI alliance. Microsoft started canceling most of their internal Claude licenses and shifting toward GitHub instead. AWS released Claude Platform on AWS, giving customers direct access to Anthropic's native Claude services. Apple reportedly is exploring ways to allow AI agents in the App Store. A new Gallup poll found that 71% of Americans oppose local data centers and prefer nuclear plants nearby instead. Open— a new OpenAI memo revealed a reorganization internally, so kind of getting some new leadership in there, unifying ChatGPT and Codex under President Greg Brockman in the new structure. A new report said that Microsoft is looking at a potential Inception acquisition. Anthropic reportedly signed a term sheet for a $30 billion funding round at a $900 billion valuation. Anthropic launched Claude for small business. OpenAI and Microsoft reportedly capped revenue sharing at $38 billion as part of their updated agreement. Meta rolled out Muse Spark conversation mode with fast voice chat, web search, and vision. In a very similar release, Probably a little more impressive, I'd say. Thinking Machines previewed real-time interaction AI models enabling continuous multimodal human collaboration. That one is pretty impressive. I haven't really seen a live model like that that can, you know, the new OpenAI GPT model, GPT Real-Time, is fairly close, but the Thinking Machines, well, we'll see when and if it comes out, but it looks pretty good. Google is reportedly hiring hundreds of engineers to help clients adapt its AI Anthropic is reportedly close to acquiring developer tools startup Stainless for $300 million, which Google and OpenAI both use. US Senators Bernie Sanders and AOC introduced bills pausing all AI data center construction, threatening billions in local jobs. OpenAI is reportedly developing a codex setting allowing Mac use while the device is locked. Ooh, I'm going to like that one. One-third of Thinking Machines Lab founders reportedly departed after their initial equity vested, according to a report. That's not good for them. Anthropic is in talks for a— oh no, we already covered that one. Sorry. All right, next, over 50 xAI researchers reportedly departed after the SpaceX acquisition, many who joined Meta and Thinking Machine Labs. Amazon killed Rufus, their AI assistant, instead launched Alexa for shopping, obviously powered by AI. And last but not least, Cursor added Teams delegation and cloud agent development environments. All right, that is a wrap for this week. Yeah, believe it or not, this is the calm before the storm. So like I said, expect a ton of AI fireworks this week. Everything with, uh, the Musk versus OpenAI, uh, deliberations starting today, Google's I/O starting tomorrow. It is going to be a hot AI summer and we're just getting warmed up. All right. And we're going to be covering it obviously every step of the way. So on Mondays we do the AI News That Matters. On Wednesdays we usually do a demo show going hands-on with AI at Work on Wednesdays. On Fridays we go over our Features Fridays, going over new AI features that have been released. And on Tuesdays and Thursdays we have rotating shows. So if you're new here, thank you for tuning in. Tuning in. Please, if you haven't already, subscribe to the podcast. It takes you like 5 seconds and it really helps us. If you could also leave us a rating, I'd appreciate that. So thank you for tuning in. I hope to 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!