EP 596: GPT-6 rumors, Elon recruited Meta for OpenAI takeover, and more AI News That Matters

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.

Google had so many AI wins this week, we could only talk about their massive movements today. And that would cover a majority of the AI news for this week. I mean, they shipped some huge AI model updates, they brought agentic features to Google search, and they may beat Amazon and Apple for the first new age AI assistant in your home. And that's just the low hanging fruit of what Google was up to. But their biggest AI competitors did not have the best week. I mean, Apple is reportedly so far behind that their revamped AI Siri may need Google's models. And OpenAI is already shifting the focus to GPT6 after a rocky rollout to GPT5. Sheesh. So much going on this week. It was actually hard to whittle down what is the most important AI news. But that's what we're here for and let's dive into it. What's going on, y'? All? My name is Jordan Molson and welcome to Everyday AI. This is your daily live stream podcast and free daily newsletter helping everyday business leaders like you and me not just keep up with AI, but how we can make sense of it all and leverage it to get ahead to grow our companies and our careers. If that's what you're trying to do, it starts here with the unedited, unscripted live stream podcast. But if you want to take it to the next level, your everyday AI dot com. Let me repeat that, your everyday AI dot com. So every single day we recap the highlights from our live stream podcast and as well as keep you up to date with everything else happening in the world of AI in that free daily newsletter. So if you haven't signed up, make sure you go do that. And with that, let's get straight into it. There's so much to cover this week in the AI news that matters for August 25th. And if you are brand new here, almost every single Monday we do this AI News that matters so you don't have to waste hours every single day wondering about what all of this AI movement means. We spend hours every single day and then we on Mondays break it down for you. So let's get into it first and foremost. Yeah, GPT6 is already a thing. That's because that's where Sam Altman has kind of shifted his focus right now. So According to reports, OpenAI CEO Sam Altman said the company is said the company totally screwed up GPT5's rollout after making it the default over their previous, apparently very beloved model GPT4.0, which triggered a swift backlash and a partial reversal that they later restored 400 for paying users. So apparently Sam Altman went out to dinner with a bunch of journalists and this is one of the topics that came up. So Sam Altman already talking about GPT6 and admitting that there was some, some problems, we'll say that, with the GPT5 rollout. So Altman also said that upgrading a product for hundreds of millions of people in one day was a mistake, noting that people had formed habits and even emotional bonds with GPT4. Oh yeah, we already went into that. It's, it's, I don't know y', all, it's kind of, it's kind of weird. But we broke that down about two weeks ago. So if you want to go listen to that episode 587 GPT 5 was canceled for being a bad therapist and why I said I thought that was a bad idea, that people are using it as one. But Altman estimated that fewer than 1% of users ultimately have unhealthy relationships with GPT4O in that regard. But when you talk about 700 million people, that's still millions of people that were probably way too reliant on OpenAI's old model to just for their day to day functioning. Also, Altman is reportedly shifting his focus to GPT6. Yay. We can already start hyping GPT6 now, even though GPT5 has been out for like not even two weeks. So the mandate for GPT6 is to feel much more personal without exploiting vulnerable users, which likely means stronger safety guardrails, more user controlled tone settings which I'd look forward to, and explicit consent for companion like behaviors. So Altman said the gap between GPT5 and GPT6 will be shorter than the jump from GPT4 to 5. But GPU shortages are a gating factor and OpenAI did say, and Sam Altman said that OpenAI had better models but can't deploy them widely due to capacity limits. He also in the meeting floated the idea that demand could require trillions of dollars in new data centers. And then we did see, if you read our Data Daily newsletter each and every week, you did see OpenAI's CFO kind of hint that they may eventually get into the infrastructure game, which would obviously make sense because right now they're having to pay technically some of their competitors in Microsoft and Google to help serve their models. All right, more OpenAI drama. We're just going to get that all out of the way first and then focus on what's new. But this one's pretty big. So OpenAI is asking a court to. To compel Meta to produce documents about any coordination with Elon Musk's X AI. So essentially, reports just came out that Elon Musk and his XAI slash Grok kind of AI company asked Meta and specifically CEO Mark Zuckerberg to join them in a bid to buy OpenAI for unsolicited bid. FYI, that we covered and destroyed this. I don't know like what this was all about, but a $97 billion offer to acquire OpenAI, which obviously is a laughable, not a serious offer. And that was back in February, which obviously OpenAI rejected. So OpenAI says that it subpoenaed Meta in June after learning that Musk reached out to Mark Zuckerberg about the XAI bid, including inquiries about potential. Potential investments, though it's unclear whether Meta holds responsive documents. So Meta objected to the subpoena in July, and OpenAI is now seeking a court order for Meta's records concerning any actual or potential restructuring or recapitalization of OpenAI, which is central to Musk's lawsuit. Yeah, this lawsuit's been going on, which in AI years, it seems like it's like 30 years it's been going on, right? Elon Musk suing OpenAI, a, a company that he kind of helped start and was on the board of, you know, being an original co founder. So a Meta spokesperson pointed to OpenAI's filing, noting that neither Meta nor Zuckerberg signed Musk's letter of intent and Meta declined further comments. And OpenAI and Musk's legal team did not immediately respond to reports. So Musk's broader lawsuit challenges OpenAI's conversion of its for profit arm into a PBC or a public benefits corporation, arguing that it violates the company's founding mission, which is not illegal, while OpenAI says the structure is needed to raise capital and potentially go public in the future. So pretty interesting now that we technically have a third party that seemingly was roped into this ordeal. So we don't know yet if these conversations were fruitful or went anywhere when Elon Musk reached out to Mark Zuckerberg. So I'm going to be curious if did Elon need the money or was he just trying to get kind of the second tier AI companies? And I'm not saying that in a bad way. Right. I think that it's very clear that OpenAI and Google are kind of on the One A1B or One B1A. Whichever way you look at it. Microsoft obviously benefiting from that partnership with OpenAI. I, I think they're right up there in the 1C. Then you have a little bit of a drop off and then you have anthropic XAI and Meta as well. So maybe it's, you know, the closest competitor is trying to team up to for this kind of takeover bid and which by the way, the, the whole bid that we covered that at length back in February, it was essentially a ploy from Elon Musk to try to stop OpenAI from growing. Right. And you know, essentially Elon Musk has a bone to pick with OpenAI over the past few years because he says that they've strayed from their original mission, which he invested heavily in in the early years. This one's juicy, y'. All. And I might have to hot take Tuesday. This one tomorrow. We'll see. So the headline and the story that everyone's been talking about over the past 72 hours or so. So a new MIT study kind of says that 95 of generative AI pilots aren't delivering positive ROI, raising concerns from some that there is an AI bubble. But companies claim they are scaling by faking fixing accuracy and workflow issues. So according to MIT's State of AI in Business 2025 report, which was based on 300 public deployments and 150 executive interviews, 40% of organizations say they've deployed AI tools. And I'm wondering like what companies are they talking to? What companies? 60% of companies haven't deployed AI tools. Like, I don't know, like when I read that, I'm like, I, like I really want to know every single company and who they talk to because I, that doesn't make sense. We're in the year 2025. There's, there's very few companies that are still around that haven't deployed AI tools anyways. Their study found that only 5% have integrated them at scale. So the study says most projects stall in pilot purgatory because employees must double check outputs, wiping out efficiency gains and eroding trust in high stakes settings. Investors are reportedly shorting AI stocks over the past few days on fears that enterprise adoption is scaling, although the report highlighting a small cohort that is successfully scaling. So the company cites adoption by Fortune 500, government agencies and regulated industries, positioning itself as evidence that the right design choices can close The Gen AI divide. Ah, so this is the news recap. So maybe I'll save this. I don't know. Should, should I do a hot Take Tuesday on this one? I kind of, last time MIT had a viral study, their brain rot, which was one of the most, in my opinion, ill conceived studies that I've read over the past few years. I don't know, say MIT yes or MIT no. All right, so I think on Spotify you can leave a comment on the show. So go ahead, do that if you want the hot take Tuesday. That's kind of what we do on Tuesdays. If I should do this live stream audience let me know. MIT yes or MIT no. Should I, should I dig into this study a little bit as a former journalist like this one, unlike MIT's brain rot study, which was their fault that they put out a terrible study, this one is not so much that MIT put out a terrible study. I would not have put out the study in this way. As someone that literally I've been reading and reporting on studies for 20 years, not the best one. So let me know. MIT yes, MIT no. Should we tear this one up a little bit for Hot Take Tuesday? All right, here's. Speaking of hot takes, here's one from Microsoft's AI boss. Our next piece of AI news. So Microsoft's Mustafa Suleiman warns of risks from AI becoming maybe sentient. So Soliman is the head of AI at Microsoft and has raised some urgent concerns about the rise of seemingly conscious AI or SC AI. All right, so when I say Scai talking about seemingly conscious AI, which are AI systems designed to convincingly imitate consciousness without actually being conscious. So he just post a very good read about this on his personal site and we did link to this in our newsletter last week. That's why you gotta read it every day. So he explained that SCAI is not science fiction, but an imminent reality achievable with current technology and simple engineering over the next two to three years. So when you combine advanced language skills, memory, goal setting and apparently self awareness, that's when you might get to this seemingly conscious AI. So the core risk arises, Solomon said, because humans are naturally. Humans naturally attribute consciousness to entities that talk, remember and express emotions, even when these are simulations leading to a growing number of people who may believe AI companions are truly sentient. So Suleiman warns this belief could lead to serious societal problems, including AI psychosis, where users from false form, false attachments or delusions about AI, they advocate for AI rights. So they treat AI like they're humans or confuse AI companionship with real relationships, which we actually saw already happening with millions of people when GPT or OpenAI essentially just wiped out GPT4O overnight. And millions of people apparently lost their screw, lost their noodles because they had formed probably some, in some cases unhealthy attachments to a specific model. So yeah, that one's gonna be interesting. Actually gonna see if we can get Mustafa on the show here in the coming months. All right, now it's, it's turning the page on the Google Hour. Like Google had so many both major and low key moves this week and some leaks and it's hard not to talk about many of them. So first Google is expanding their AI mode with a J ticket features. So they just announced this at their big event on on Friday. So these features are set to launch first for Google AI Ultra subscribers in the US Though the agentic capabilities in AI mode is an experiment in labs, making it limited to early access release. But here's a little bit about what will eventually be coming on the agentic side to Google search and Google CEO Sundar Pichai did say, he did say previously that Google's normal search would eventually be replaced by AI mode, which is why these updates to AI mode are really important to talk about. So now AI mode also is expanding to over 180 additional countries and territories in English and that is already kicked off. So you don't have to have a paid Google plan to have the Google AI mode. But these agentic features that are rolling out at least right now, they're limited to those paying the hefty price for Ultra. So the new experience pulls real time availability across reservation platforms and presents a curated list of restaurants that match specific constraints such as party size, date, location, cuisine, then links directly to the booking page to complete the reservation. So yeah, if you give Google AI mode a little bit of criteria, it'll actually just go agentically do the reservations for you. So under the hood, Google said it uses live web browsing via Project Mariner, which I've enjoyed using, and direct partner integrations and signals from their Knowledge Graph and Google Maps partners right now include OpenTable, Resi Tock, Ticketmaster, StubHub, SeatGeek, Booksy and others. So right now the agentic capabilities available in AI mode are mainly focused on reservations and buying tickets, although there are plans for this to roll out more broadly across different services and businesses in the near future. Also, there's some new personalization updates, so those features are live for US users opted into AI mode in labs and that surfaces dining results tailored to inferred preferences according to what you search. So users also do remain in control of personalization with settings adjustable in their Google account. There's also a new link sharing feature which seems pretty cool, that lets users share share AI mode results so friends and family can continue the same thread. So yeah, if you haven't used AI mode, it's actually really, really cool. And I think it's a cool way for people to experience a little bit of AI. I'm always, you know, telling my wife when she's googling something on her phone, I'm like, oh, click on AI mode and keep talking with it. Right? So I think it's just a nice way to get kind of almost like deep research esque, because even in AI mode, the response might only take five seconds, maybe less, three seconds, and oftentimes it's going to 50, 70, 80 websites each and every time. So Google obviously has a huge advantage there with their cache of pages that they can go through very quickly. More Google y'. All. So. Are you still running in circles trying to figure out how to actually grow your business with AI? Maybe your company has been tinkering with large language models for a year or more, but can't really get traction to find ROI on gen AI. Hey, this is Jordan Wilson, host of this very podcast. Companies like Adobe, Microsoft and Nvidia have partnered with us because they trust our expertise in educating the masses around generative AI to get ahead of and some of the most innovative companies in the country hire us to help with their AI strategy and to train hundreds of their employees on how to use Gen AI. So whether you're looking for ChatGPT training for thousands or just need help building your front end AI strategy, you can partner with us too. Just like some of the biggest companies in the world do. Go to your everydayai.com partner to get in contact with our team or you can just click on the partner section of our website will help you stop running in those AI circles and help get your team ahead and build a straight path to ROI on Gene. Are we finally going to get some actual smart AI assistance? Well, maybe. Google has unveiled its plans for Gemini for Home to replace their Google assistance. So Google is rolling out an all new voice assistant for the home in October, at least for early access, signaling a major shift in smart home control and a direct challenge to Amazon's Alexa plus, which, I don't know, does anyone actually have access to that? So anyways, Google says their new Gemini for Home will ultimately replace their older AI assistant. So if you look at the, I like to say the dumb AI assistants that have been around for 10 years, right, the original Siri, original Alexa, original Google Assistant, and now we're ushering in this wave of large language model power word AI assistant. So that's what we're seeing here. Google getting official, talking about their Gemini powered Google for home. So now it kind of stays on in the background and some of the first offerings are apparently going to be rolling out on Nest speakers and displays so you can keep the kind of hey Google wake phrase while adding more natural context aware voice control. So the rollout begins in October with an early access program for both free and paid users, though Google has not announced pricing for the premium plan. So Gemini Live will also be integrated for more conversational back and forth without having to repeatedly say hey Google or having to repeat the context, right? That's one of the things that's worse for me. Try to use these old school dumb, you know, voice assistants is, you know, for three plus years, right? Using large language models even before that, but more in the chat interface, right? You're so used to having that context so you know, having 5, 10, 15, 20 conversations without having to repeat the context or repeat what you're talking about. And that's when you're trying to use these home assistants. It's, it's terrible, right, because we've ingrained probably a good habit of the smart AI, you know, kind of remembering things. So pretty exciting here. So Google may bundle some premium features into existing subscriptions such as Nest Aware or Google One. So access will likely be gated through Google's public preview program at first with a cautious phased rollout due to safety concerns around controlling critical devices like locks, appliances and H vac systems. So yeah, I might have to, you know, get, get myself one of these. I have a couple Alexas around the house, but I, I haven't gotten access to the Alexa plus which is powered by Claude. So I, I don't know, might have to jump ship there with a smart home assistant because chances are we're going to be waiting until like 2045 for Apple to roll one out. Speaking of Apple and voice assistants, talk about an intentional transition. So Apple is so far behind in this Siri that they've been promising for like a year and a half and they face multiple lawsuits, class action lawsuits about this because they're like, oh, buy our new hardware, you're going to get this new smart AI. Siri, you're not they can't build it and it's getting so bad. We've reported on this previously that they were maybe going to partner with OpenAI but according to a new report from Bloomberg, they're actually maybe going to be partnering with their biggest rival in Google. So Apple, according to Bloomberg report, is in early talks to use Google's Gemini as the backbone for their new version of Siri. So according to the report, Apple has approached Google to build a custom AI model for Siri that could debut next year. With Google, Google already training a version designed to run on Apple servers. That is huge. The fact that they've already started training a version, I mean Gemini's voice assistant, Gemini Live, really, really good. So if they've already are trying to train a version specifically for Apple which you know, has to run on their, you know, private and semi private cloud. I don't understand Apple's cloud, but that's pretty big news right there. So the discussions are described as early and private, but meaning terms could change or the deal could fall through. But the talks signal Apple's willingness to source core AI tech externally and especially from one of their biggest competitors on the hardware side. That's the part that's crazy to me. Yeah, it didn't really take me too far back or it didn't take, I wasn't too far taken aback by previous reporting that we've talked about on the show that Apple was reportedly in talks with OpenAI to use their kind of advanced voice mode AI powered tech because they have the partnership right now. So with the Google one, this is a little surprising. So running the model on Apple's servers according to report, does suggest more of a hybrid approach because Apple can enforce privacy and security and also control how Google's AI is leveraged. And this is pretty big news because there would probably be some little regulatory scrutiny. Yeah, basically any deal that could come between Apple and Google, two of the five biggest companies in the world or two of the four biggest companies in the world is going to be heavily scrutinized by the Department of Justice, potential antitrust attention. So yeah, it could help benefit anyone that is an iPhone user, that's myself. But we'll see if it actually, the deal actually happens or if maybe it falls through. All right, we have a one, one slide I forgot to put up. All right, so I don't, I don't have the slide for this one. Sorry y'. All. But this one is pretty big news. So according to a new report, Meta is freezing hiring in their AI division amid Some market jitters. So Meta has frozen all of its hiring in its AI division, part of a leadership reshuffle, according to a report from the Wall Street Journal. So the move comes as tech. As tech stocks late last week tied to AI saw a sharp sell off with shares of Nvidia and Palantir dropping, fueling concerns that valuations ran ahead of current capabilities and revenue. So yeah, we saw these stories. Oh, Meta is pausing AI hiring. That doesn't make sense. They've spent billions, literally billions of dollars acquiring like a dozen people, right? Giving them a lot of them multi hundred million dollar offers over the course of a few years. So yeah, for a lot of people this isn't adding up. This story also Alexander Wang, who's met us, recently hired chief AI officer, and you could say he might be like, in theory at least, when it comes to AI, he's second in command, right? Aside from CEO Mark Zuckerberg. So on Twitter, Wang denied the reporting, saying, and I quote, we are truly only investing more and more into Meta Super Intelligence Labs as a company. Any reporting to the contrary of that is clearly mistaken. All right, so let's look a little bit more on this report. So Meta reportedly characterized the freeze as basic organizational planning, aligning it with a broader restructuring of its AI leadership and strategy, rather than a budget cutting retreat. So as part of this shakeup, Meta's Super Intelligence Lab, or msl, will now house four units. This is like the third time they've reshuffled their AI organization in the last few weeks. So I don't know, maybe we should hand the just the organization part over to AI so we don't have to change it every week and get our audience confused, including myself. I'm reading these stories, I'm like, didn't I report on this the last three weeks? Yes, I did, and they keep shuffling it around. Anyways, so now you have Meta's superintelligence lab housing four units total. So MSL is the umbrella company. So then you have the. The four different units and the AGI Foundations Group responsible for the latest Llama models. That one has been dissolved after the last rollout of their models kind of drew some criticism over some alleged benchmark inflation. So at least three members of the AGI Foundation's team have announced departures, according to the Wall Street Journal, raising questions about continuity on Meta's core model work. But before the reported freeze, Meta hired more than 50 people for AI, including roughly 20 researchers and engineers from OpenAI, many of them scoring multi million or 100 million dollar plus deals. So because of this and a few other things that happened this week, including, you know, OpenAI CEO Sam Altman referencing in AI bubble analysts have raised red flags about Meta's reported billion dollar scale compensation offers for key recruits, warning that oversized stock packages could dilute returns for shareholders if performance lags. But the spending backdrops remains intense across industries, with firms investing tens of billions in AI infrastructure while revenue trails increasing the risk that costs may outpace near term payoff for companies and investors. So Mark Zuckerberg insists the new organization can deliver on ambitious goals and including automating all valuable work. What he said but execution risk is high given leadership churn product credibility questions and market skepticism. And I don't know, is everyone going to remember what team they're on on Meta when we keep reshuffling the organizational structure and what everyone works on. So hopefully this is the last time that we have to talk about, you know, MSL restructuring and all their on different teams underneath. Right? But all right, that's a lot. But there's still a lot more. So let's talk about what's new and what's next. So this is kind of our new ish. I think we've been doing this for a couple of months. There was so much news this week we couldn't fit it all and give everything an actual story. So this is more of our bullet point roundup of things that were either just released, some things that have been rumored, some things that are in testing right now. There is an absolute insane amount. Ready. All right, first, Meta and Mid Journey are reportedly teaming up and plans for future licensing deals in Meta's products. So Meta may be getting some better AI visuals. Google teased a Gemini enabled smart speaker at their Made by Google event. It was just kind of in the background and everyone's like wait, that's a new Google speaker. Yeah. Speaking of Google teasing things, they've been teasing very a lot. They've been teasing Nano Banana, their AI image generating model that could shake up the entire creative industry. I don't want to be Adobe when Nano Banana drops and when it drops on the API. My gosh. We've been sharing about this in the newsletter, but maybe we'll do an actual show when it drops and maybe get someone from the from that team on the show as well. If you guys would like that, let me know. Next, we have a new coding model kind of under the radar that was spotted in Cursor called Sonic. Some are saying it might be Grok's new coding model Google may be bringing its deep research to Notebook LM According to testing catalog, a new Google Docs is has rolled out AI generated audio for documents in Google Docs. Google Photos bringing conversational AI editing to Google Photos with the Pixel, their new flagship device. The Pixel 10 getting first access. So pretty cool. Like AI photo editing just with your voice. Deepseek has released its newest model, Deepseek V31. Still no updates yet on their R2 model. 11 labs released their text only AI agents. Microsoft is rolling out a new Copilot function in Excel beta powered by GPT4.1. OpenAI this is a big one. This is how much like news there was. OpenAI is rolling out Project Memory, which seems like a very small thing, but it's actually huge and we're probably going to be rolling this out into our finally refreshed prime prompt Polish V2 course. So project Memory, really, really big Perplexity. Speaking of memory, Perplexity is working on and teasing what they're calling Super Memory for users. As their CEO said that memory will ultimately be the moat for those companies using AI. Google is testing a way to export Gemini Canvas Creations to its Firebase Studio, which will be amazing. I can't wait for that one to happen. And OpenAI added Gmail, Google Calendar and Dropbox support to its Responses API, which you might not really. I mean unless you're a dev, you might not understand what that means, but essentially a whole bunch of AI tools out there are now going to be working better and more agentically with your Gmail and Google Calendar. So if you are a Google organization versus a Microsoft in terms of your email, calendar, et cetera, that one's huge. There's going to be probably a ton, I would guess dozens of pretty decent startups that are either launching or updating their current projects that are going to just make it easier and better to work through your email and your calendar. So a lot of your day to day things, it might actually be working much better now and being able to take action on those things. So small update it seems, but big news. That's what's next and what's new. My gosh, we got to a lot. Let's very quickly recap what was new this week. There's a ton. So CEO Sam Altman admitted that GPT5 got off to a rocky start and the backlash is actually shaping plans for GPT6 as he's already talking about it. OpenAI is seeking records from Meta after Elon Musk's apparent bid to get Mark Zuckerberg to join XAI in a OpenAI takeover. Juicy the new MIT report says that 95% of AI projects failed. Microsoft's Mustafa Suleiman warns of AI risk from seemingly conscious AI. Google expanded their AI mode with some agentic features and rolled it out to more countries. Meta is reportedly freezing AI hiring, even though its chief AI officer said not true. Google unveiled Gemini for Home to replace their old AI assistant, and they might be the first big company to bring an actual smart LLM powered AI assistant to the home and beats Amazon. Speaking of that, Apple is exploring Google Gemini's help for a Siri revamp and that's it. So much. I hope this was helpful. Don't spend hours every single day trying to keep up with the AI news and what it means for you, your company, your department. Join us on Mondays. This is what we do. We've been doing this every single Monday for our year and a half at least, so I hope this was helpful. Join us tomorrow for our hot Take Tuesday. On Wednesdays we do putting AI to work at Wednesdays. We'll probably put a poll in our newsletter either today or tomorrow asking you all what you want for that. I think that's been a pretty well received segment. On Wednesdays, people are always like, oh, so many AI tools and modes. What do I do? Well, we get to work with you live on Wednesday, so make sure to join that. And then on Thursdays and Fridays, you know, usually we do interviews, bring a bunch of smart people on and talk about certain topics in AI. So if you haven't already, please share this with someone. Don't keep it to yourself. If you're listening on the podcast, please leave us a rating and subscribe to the show. Then go to your everydayai.com Sign up for the free daily newsletter. Thank you for tuning in. 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 your everydayai.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.