Claude Opus 4.5 drops, U.S. government makes bold AI move and ChatGPT ads landing soon? 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.

A week after OpenAI, Grok and Google all released their most powerful AI models to date, Anthropic followed suit with Claude Opus 4.5. But that might not even be the biggest AI news of the week. That's because OpenAI was busy making headlines on AI hardware and potentially ads in ChatGPT, while even the federal government is throwing its hat in the ring on trying to create AI breakthroughs. So maybe you were in a little bit of a turkey coma this week over Thanksgiving holiday and you missed some of these AI news headlines. Don't worry, I'm going to be getting you caught up today on Everyday AI. What's going on y'? All? Welcome to Everyday AI. My name is Jordan Wilson and this is your daily live stream podcast and free daily newsletter helping everyday business leaders like you and me not just keep up with these AI news headlines, but how we can make sense of them and use this information to grow our companies and our careers. So if that's what you're trying to do, awesome starts here with the unedited, unscripted live stream podcast. But to take it to the next level, make sure you go to your everyday AI dot com. Go sign up for today's free daily newsletter. We're going to be not just recapping today's show, going over the AI news, but keeping you up to date with everything else that's happening today. So if you are brand new here, maybe listening for the first time. Most Mondays we do this AI News that Matters segment so you don't have to waste hours every single day trying to play with new tools or trying to see know who dropped what model. That's what we do. So join us on Mondays as we bring you the AI news that matters most. Tuesdays we do hot take Tuesday. Wednesday we do AI at work on Wednesdays. Kind of a hand on hands on kind of demo. And then usually Thursday and Fridays we do interviews. So let's get into the big AI news that matters this week. If I sound a little hoarse, it's because I am feeling a little bit under the weather. But AI news doesn't take a day off and well, I guess I don't either. All right, first, hardware OpenAI. Yeah. So OpenAI CEO Sam Altman says OpenAI is finally finished its first AI hardware prototypes and called them jaw dropping Good in a discussion hosted by Emerson Collective and reported by CNBC. So reportedly the new AI prototypes from OpenAI are the first big steps since OpenAI bought Johnny I've's startup IO for $6.4 billion in equity back in May, signaling OpenAI's serious push into AI powered consumer devices. Altman would not say what the actual device is, just that it was jaw dropping good, but said it aims for a calmer experience than smartphones. Comparing phones to walking through noisy Times Square and his vision to quietly sitting in a cabin by lake, Altman said the device would look over long periods, filter out distractions and only notified users what really matters? Potentially knowing everything you've ever thought about, read and or said, which obviously raises a lot of personalization and privacy trade offs and well, maybe more on some of this data collection here later in the show. So Johnny Ives, the former Apple design chief is now leading the hardware push for OpenAI said he expects the device to be revealed within two years, while warning the hardware development is unpredictable. So according to reports, the project is being watched closely because a successful OpenAI device could pressure Apple, which has delayed major Siri upgrades to 2026 and has yet to launch an AI focused device design. OpenAI also recently announced a deal with Foxconn, which is Apple's major hardware supplier, to build AI infrastructure in the US which deepens its hardware t device. So it's going to be interesting what this device is. A lot of people said it's, it's, you know, calling it a third device. So it's not a phone, it's not a laptop. A lot of people have speculated that it's more of a, you know, small puck type thing that you would put into your pocket and it would just hear everything that goes on. What's interesting here is kind of this nod to also anything you've ever thought about, right? So obviously I don't think most people are assuming that this is a neural link type device, but that's interesting. So maybe it's just something a small thing that you would put in your pocket and you know, kind of records every single conversation that goes on, maybe. Which would obviously have a lot of privacy implications or something that you could just, you know, talk to at any time. Right. So almost like a chat GPT in your pocket, but you can just, you know, talk through problems of your day, etc. Save, you know, to DOS for later. Not sure. But regardless, we've covered on the show a lot over the past couple of weeks. The amount of hires that OpenAI specifically has kind of swiped away from Apple reports have said Dozens of Apple AI hardware leaders have left Apple and have joined OpenAI. So regardless, it's whatever it is, it's either going to be, I think, a huge hit or a huge flop. Right? OpenAI obviously has fundamentally changed the way that the everyday person interacts with technology because of ChatGPT and large language models. So I would assume that OpenAI is going for a big swing here with whatever their AI hardware device is. Speaking of big swings, the federal government is making one. So US President Donald Trump has signed a new executive order creating the Genesis mission, a large scale federal push from the US to use AI and supercomputers specifically to speed up scientific discovery. So this new initiative directs the Department of Energy and its 17 national labs to build an integrated AI platform that connects their current and future supercomputers, aiming to cut research timelines in areas like health, energy and engineering from years to days, or maybe even hours. So this new centralized AI platform is designed to analyze massive federal, university and private sector data sets and to automate tasks like experiment design, advanced simulations and predictive modeling. So the administration says the most visible impact could come in healthcare, where faster AI driven research could help develop new therapies for serious diseases that are currently often fatal, potentially changing treatment options within the next five years. The White House says private companies including Nvidia and Dell have already shown interest in contributing AI supercomputing capacity, indicating that this mission could further tighten the link between government labs and leading tech firms. Yeah, we've already seen that line start to blur a little bit, even with the, the U.S. government taking a equity share in a public tech company. The first time I believe that's ever happened. So another kind of intersection here. But, you know, when I read this last week and we share in her newsletter, I was kind of scratching my head. I was like, okay, this is good, but why had this not even existed before? Right. And maybe it's because, you know, the US doesn't really have a kind of large AI presence, presence across different branches of government. So maybe this is looking at using the Department of Energy specifically. Well, because for AI, compute and infrastructure, well, you need energy. So it should, I think it. Sorry, it should, it shouldn't go without saying that this is extremely important for the U.S. right. So we're going to be talking about this probably next month when we go over our 2026 AI prediction and roadmap series. Uh, however, sovereign AI is going to be definitely at the top of most people's lists as you look at 2026 and what's happening in AI and what that essentially equates to is, well, governments specifically looking at AI as more than a technology and really as a natural resource and probably one of the most important kind of resources that any country could have. In the same way that, you know, they might view oil or gold or even their military, I think the a. I think AI is probably going to meet or exceed that for most countries in 2026. So I'd say this is a pretty good move and a good first step from the US in this genesis mission to kind of centralize AI, compute and information sharing across different branches of government. All right, Nvidia taking shots. Yeah, didn't see this coming. So Nvidia making headlines this week as it went out of its way to say that its AI technology is still a generation ahead of the industry, pushing back on fears that Google's in house AI chips could chip away at Nvidia's dominance. So, yeah, this statement that they put out online, which a lot of people were scratching their heads about, followed a roughly 3% drop in Nvidia's share price on Tuesday after reports that Meta may use Google's TPUs, or tensor processing Units, in its data centers instead of relying solely on Nvidia's GPUs. So in a post on Twitter, or X, as some people call it, but I refuse to, Nvidia said that it is, quote, unquote, delighted by Google's success and noted that it continues to supply chips to Google while also arguing its platform is the only one that can run all major AI models everywhere computing is done. So, yeah, Nvidia essentially has contrasted its GPUs with specialized chips like Google's TPUs, claiming their GPUs offer better performance and flexibility because they can run many types of AI workloads. Workloads for many customers, not just a single company or function. Whereas a lot of companies now are looking at Google's TPUs for inference, right? Where I think with Nvidia, they still have an absolute stranglehold on using GPUs for training. So analysts estimate that Nvidia still controls more than 90% of the AI chip market. But Google's TPUs are drawing fresh attention as an alternative, especially as Nvidia's newest Blackwell GPUs are both extremely powerful and very, very expensive. So Google does not yet sell its TPUs directly. Instead, it uses them internally and rents TPU capacity to customers through Google Cloud, giving companies another way to run large AI models without buying their own hardware. And this obviously comes not just the meta news. Reportedly entering into a partnership to use Google's TPUs, but this also comes right after Google Google launched its very successful and extremely capable Gemini 3AI model. So essentially there's been a lot of industry hype and I would say it's well deserved around not just Google's AI models in Gemini 3, but also how they've been able to train their models with TPUs. So Nvidia took a rare step this week and literally just putting out a tweet that was like, all right, we see you Google, but we're still far ahead of everyone. So pretty, pretty interesting kind of PR or comms move there from Nvidia.

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.

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, or 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. We'll help you stop running in those AI circles and help get your team ahead and build a straight path to ROI on Gen AI.

All right, speaking of interesting moves, maybe another one we didn't see coming but could save you a little bit of time. So OpenAI has announced and has already rolled out a new shopping research feature in ChatGPT that turns any shopping related question into a guided product finding session. So the feature is available on all ChatGPT plans, both free and paid and mobile and web with what OpenAI says is nearly unlimited usage during the holiday shopping season, making making it likely to reach a very large audience quickly. So as an example, if you ask ChatGPT what's the best TV in bright lighting, ChatGPT now responds with normal text but then adds a button offering to do additional research which starts a more structured personal shopper style experience. Or if you've used deep research before with OpenAI and inside ChatGPT. It's kind of like a slightly fast, faster, but more interactive version of that. So the new shopping flow asks follow up questions about budget, intended use in this example, and specific features working much like filtering menus on major retail sites, but inside of a chat instead of a web form, which could save time for busy shoppers. OpenAI says the new system is powered by a version of GPT5 mini tuned specifically for shopping tasks, using quality sources across the Internet to surface product details such as current prices, availability, reviews, tech specs and images. The company does warn that Shopping Research can still make mistakes on key details like price and stock levels, and urges users, well, to be the human in the loop. Products shown in ChatGPT can be purchased on external retailers sites, and OpenAI does plan to add an instant checkout feature so that users can buy certain items directly inside of ChatGPT from participating merchants, which could further obviously blur the line between search advice and purchasing or ads when and if technically OpenAI does roll out those ads. So I did use this a little bit over the weekend and I was kind of lukewarm on it. I don't know, I think using a the GPT51 Mini, right? Even comparing, right. I did a quick comparison of a kind of a shopping style or shopping related query using normal Deep Research, which presumably uses the GPT 5.1 thinking, and then using this kind of new shopping research assistant which uses GPT5.1 or a version of GPT 5.1 Mini and the GPT 5.1 Mini, at least in my very limited testing, let me say that with a huge grain of salt. It was bad, right? Again, compared to using Google's or sorry, compared to using OpenAI's deep research, which Deep Research is extremely, extremely good or even using. Similarly, did just did a normal query using GPT 5.1 thinking, you know, induce the thinking up a little bit. So I don't know, maybe for free users there might be some utility in this. Obviously I think this in the long run is a way for OpenAI to monetize through affiliate partnerships, especially when they start bringing more of these purchase options available in app, right. I think that's ultimately where, you know, OpenAI is probably going to be making much more money on the commission or affiliate side. I don't know. But first impressions for me and very limited testing, you know what I did is I asked what, you know, Black Friday deals there were for different AI software, right? And the shopping version came back with a bunch of irrelevant things. It was trying to sell me like prompt packs, right? So it did a very poor job. But you know, doing it obviously in the normal deep research or using GPT 5.1 thinking gave me much better results. So I don't know, maybe for free users this could help, you know, if you're asking more about specific products, but for me, probably not going to be using it anytime soon. All right, next Amazon will invest up to $50 billion to expand AI and supercomputing infrastructure for US government customers on AWS, according to well Amazon, so this is one of the largest ever public sector cloud commitments to date and shows how central secure AI computing has become to both big tech and federal agencies. So starting in 2026, the project will add nearly 1.3 gigawatts of AI and high performance computing capacity across AWS top secret AWS secret and AWS Gov cloud regions. So well, if you're like me and you're like what the heck is a gigawatt? Well, you know, so this project will add 1.3 gigawatts so 1 gigawatt of power can support about 750,000 US households. Meaning this expansion represents the rough equivalent of the power demands of close to a million homes focused on government AI workloads. So federal agencies through this partnership will get access to AWS services such as Amazon SageMaker for training and customizing models, Amazon Bedrock for deploying models, and AI agents and foundation models, including Amazon's own nova and anthropic. Claude so AWS's CEO Matt Garmin said the initiative aims to remove tech barriers that have held governments back, potentially enabling more automation, efficiency and cost savings in how agencies operate. So analysts note that while Amazon still leads the cloud market, its lead is pretty slim, especially with AI related cloud growth from rivals like Google and Oracle. Also, this investment reflects the broader AI race between the United States and China as Washington seeks to expand expand its AI compute capacity to maintain a strategic edge. All right, mentioned Claude in there and that is our next piece of AI news. So Anthropic, kind of just low key released a pretty big model. So Anthropic released its most advanced AI model, Claude Opus 4.5 this week, which scored higher and this one's pretty noteworthy than any human candidate that they've ever had on their own internal tough engineering assessment. So the new model also comes at a much reduced costs compared to even Sonnet 4.5 or Opus 4.1. So Claude Opus 4.5 is priced at $5 per million input tokens and $25 per million output tokens, a dramatic reduction from previous rates. Any move expected to pressure competitors on both cost and performance benchmarks. Really good. So Opus 4.5 achieved an 80.9% accuracy on Swebench Verified outperforming OpenAI's GPT 5.1 Codex Max and Google Gemini's Gemini 3 Pro in real world software engineering tests. So early testers report the model demonstrates improved judgment and intuition, handling complex workflows and producing coherent summaries from internal documents and slack channels. Efficiency gains also significant. So opus 4.5 matches or exceeds prior models performance while using up to 76% fewer output tokens, helping enterprise cut costs and speed up development. Anthropic for developers on the API side also introduced an effort parameter that lets users adjust the model's computational intensity, balancing speed cost in output for different tasks. So obviously dropped some of the API information there. We did on Wednesday of Last week, episode 662, if you want to go check that out. We gave kind of our, you know, non dev, non technical review of Claude Opus 4.5. Very topical by the way. Right. I don't know for me wasn't super impressed. It could have just been, you know, tested it a little bit too early. Most of the little demos that we ran were super buggy. However, I do expect most of that to improve and I do think that, you know, I think what happened, right, If I'm just taking a look at the landscape, right? You have your four big competitors that I always talk about, right? You have Microsoft Copilot, you have Google gemini, you have OpenAI's ChatGPT and then Anthropics Claude. Right. So right now at least Microsoft is not really focusing on their own models, mainly using OpenAI and a little bit of Anthropic in their cross their Copilot suite. So you really have three frontier AI building companies, right? Sorry, no one uses Grox xai. Right, no one uses that. Mistral's great. There's other great models, but for the most part it's, it's a three model race, right? Synthropic, Google and OpenAI. I think OpenAI's base model, the new 5.1 update, at least for me and a lot of others, it's really annoying to talk to now, right? It's, it's, it's overly verbose. It's just too, too chatty. Right. I think they were trying to kind of appease the people who were upset at GPT4O going away. So I think Gemini3Pro came in as a, you know, great daily driver. Even for me personally, I'll probably still use Gemini 3 Pro as a daily DR, but use GPT 5.1 Pro for, you know, kind of the heavier lifting. But Opus 4.5 again, didn't do great in our live demos, but I think for anything coding, software development, anything on the financial side, they have a new, improved integration with Excel. I think Opus 4.5 for specialty use cases is always going to have its case. For me, not so much for a lot of, I think teams and companies, not so much. But if you are software dev, if you're in finance, if you're in Excel all day, I think Claude, Opus 4.5 is still going to be a model that you have to look at. All right, in our last big AI news story of the week. Yeah, ads are apparently coming to ChatGPT, maybe sooner than a lot of people thought. And going back to our first AI news story, right, Talking about an AI hardware device that will just, you know, see and hear and understand everything that happens in your life. Oh, that's some valuable data. Right. Okay, well, where's the. How does OpenAI monetize that? Well, here you go. Ads. So OpenAI is reportedly rolling out a new way for users to see ads. So OpenAI's ChatGPT app appears set to introduce ads for free users, at least based on a new code reference found in OpenAI's Android beta version by AIPRM engineer Tibor Blaho. So yeah, essentially this was a code leak found in the newest version of OpenAI's Android beta. And the code mentions include different features to ads, including ads Features Bizarre content and search ads carousel signaling a shift toward in app advertising. So this marks a major change. Well, if and when this happens, we'll mark a major change for OpenAI and ChatGPT, which has previously focused on paid subscriptions and API access for revenue. The move comes as OpenAI faces financial pressure. Forbes reports the company needs an additional $207 billion to fund its growth and is not expected to be profitable until 2030, according to HSBC. So CEO Sam Altman recently acknowledged on a podcast that advertising is being considered, citing Instagram ads as an example he personally finds effective. Also, you got to talk to sheer numbers. ChatGPT currently boasts around 800 million weekly active users, making it, well, one of the biggest online platforms in the world. And out of all of them, really the only one that isn't running ads, right? Any of those biggest website in the world. They're almost all monetizing on free users in one way, shape or form. So if implemented, users of the free version of Chat GPT could start seeing sponsored products in deals similar to ads that you might see on Google or Instagram. Also, we did cover in our newsletter. Google is also starting to test ads in its AI results. We've already seen that with Microsoft, so nothing new there, but it's kind of been this hot topic with OpenAI and when and if the, you know, the leading AI company would do something like that. And I think that there's a big conversation to be had around this, mainly when you're using something like Google or even on, you know, meta's platforms like Facebook or Instagram, I think you're only giving those platforms so much. In the case of Google, as an example, you're putting in search terms, you know, for Facebook, you know, maybe you're in, you know, entering your interests on a Facebook profile or you're liking or commenting on certain type of content. So meta can start to connect the dots there. Right? A large language model, specifically something like ChatGPT that many people, I will say this, unfortunately use as a therapist because I don't think that's a good idea. But that's one of the biggest use cases for ChatGPT is using it as a therapist, as a life coach, you know, as a mentor, as a friend. Right. So obviously, obviously people are revealing so much more about themselves that they would never reveal, you know, to Google or put, you know, that kind of information inside of a social network that Meta might own. So the, the ceiling for advertising is obviously huge, just from a strictly financial perspective. Right, Obviously Google has been the leader in advertising for decades, Meta as well, right. For at least 10 plus years. So we did do an entire episode last month actually. No, sorry, it was the end of October, so just over a month ago. So go back and listen to that. It's episode 641. Chat GBT ads. Nine reasons why personalized ads are coming to Chat GBT soon. I laid out right before any of this reporting, I laid out the case pretty cut and dry, right, Saying hey, yeah, ads are coming, here's why, here's what they're going to look like and here's how this will change business. So if you are interested in that, make sure to go listen to episode 641. All right, that is a wrap for the big AI news stories. Now let's get over kind of our bullet point roundup section. You know, not everything was a top AI news story of the week. There were some smaller features and some rumors and rumblings of what's around the corner. So let's kind of round up what's new and what's next. So ChatGPT updated their live voice mode for a more interactive experience. Pretty cool. I used it a couple times. I like it. Google rolled out Remix in Photos, which uses Nano Banana to remix images. Google also rolled out interactive images in Gemini which looks really cool. Black Forest Labs released Flux Point 2. I think it's just Flux 2 actually. I think it might be a top image model along with Nano Banana Pro once all the benchmarks are in perplexity with a pretty big under the radar update. They rolled out memory capabilities for its Comet data browser, putting it more in line with ChatGPT's Atlas. ChatGPT may be rolling out a new faster style driven version of their image gen called Image Gen V2, which could also include cameo style features like what's made their Sora app and their Sora AI video generator extremely popular. Being able to kind of do these cameos of your friends or people public public figures. So that may be rolling out in the image gen. So OpenAI try to keep pace with Google's Nano Banana there anthropic aside from Opus 4.5, they expanded access for cloud code desktop, Claude for Excel and Claude for Chrome. Google may be linking NoteBookLM notebooks directly in Gemini. That'll be great. Something I've been asking for for more than a year. DeepSeek released a new model called Deep Seq Math V2 that achieved gold level performance on the IMO. So pretty rare company they're in I believe just with OpenAI and Google that have achieved the same with large language models. Also Google is working on revamping the user interface in Gemini which could be a big move for them to continue to compete with ChatGPT on the consumer side. Amazon is I believe starting today hosting their Re Invent conference, so there may be some AI announcements coming soon. Meta is reportedly weighing using Google's TPUs which we already referenced earlier and OpenAI is previewing an app store like experience for ChatGPT apps. All right, that was a ton of AI news. I hope this show was helpful. Sorry, I'm a little hoarse. Hopefully I'll be feeling better soon, so if you haven't already, please share this with someone. Right? Don't let everyday AI be your little cheat code. If you're listening on the podcast, appreciate your support as always. Please if you could do me a little favor. Takes five seconds. Go ahead and click like or subscribe right, depending on if you're listening on Apple Podcasts or on Spotify. And if you could leave us a rating, we'd really appreciate that. Then go to our website, your everyday AI.com we're going to be recapping everything we talked about today and everything else that's happening in the AI world because there's always a ton. So 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.