Meta buys AI agent giant, Grok under fire for explicit images of minors, OpenAI building pens & 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.

We're not even a full week into 2026 and there's already been a huge multi billion dollar acquisition. OpenAI is trying to bring AI to your pen and by Pinterest and we have an AI product that's generating a billion dollars a year that is being built by itself. It's already been a weird 2026 in the AI world and if you miss anything, if you're still waking up from your long holiday break, don't worry, we're going to get you quickly caught up with all of the AI news that matters. Let's get into it. What's going on y'? All? Welcome to Everyday AI. My name is Jordan Wilson and this thing's for you. It's a daily livestream, podcast and free daily newsletter helping everyday business leaders like you and me keep up with everything that's happening in the world of AI. How we can leverage the most important insights to grow our companies and our careers. So most Mondays we bring you the AI news that matters. That's where we recap a crazy, usually crazy busy week of AI news and developments and tell you, hey, here's what's matters. So if you're new here, thanks for joining. We, like I said, we do this most Mondays, but every day we bring it to you live 7:30am Central Standard Time. So if you haven't already, make sure you go to our website at your everyday AI.com Go sign up for our free daily newsletter. We're going to be recapping the the highlights from today's show as well as bringing you the freshest AI developments. And hey, a little announcement later this week. Finally, after working on this for more than two years and flip flopping more than a pair of sandals on the beach, we are finally launching our AI community, the Inner Circle as well as our updated Prime Prompt Polish course. So. So keep an eye out. I think we're probably going to start letting people in on Wednesday. Everything is ready to go. I'm super excited. All right, enough of the house cleaning and announcements. Let's get into the AI news that matters for this week. January 5, 2026. First of all. Yeah, a multibillion dollar acquisition from one of the big players. That's because Meta has agreed to acquire Meta Manus, a Singapore based AI agent startup, in a deal reportedly valued at more than $2 billion. According to the Wall Street Journal making this one of the first US tech acquisitions of a US company with Chinese roots. So man, as you're like, weren't they kind of just raising a bunch of money? Yeah, they were. They're in the middle of raising new funding at a $2 billion valuation when Meta approached the company signal urgency to secure advanced AI products rather than build everything internally. So Manus is really known as a general use AI agent startup that can generate in depth research reports and build custom websites with minimal human input using models from firms including Anthropic and China's Alibaba. Interest in Manus surge after a March demo which came soon after the debut of Deep Seq, a Chinese developed AI model that drew attention in Silicon Valley for its strong performance and claimed efficiency gains, even though those efficiency gains from Deep Seq weren't true. But great timing for Manus and they've been on fire ever since the last few quarters. And Manus co founder and CEO Zhao Hong, who goes by Red, will report to Meta Chief Operating Officer Javier Alivan, placing the team directly inside Meta's senior leadership structure. So according to the Wall Street Journal, acquisition is part of Meta's broader effort to compete with Google, Microsoft and OpenAI in AI agents, a fast moving, pro moving product category designed to handle complex tasks with little supervision. So I've been saying it and I'll continue to say it, we're going to see a ton of acquisitions in 2026. This one not necessarily surprising that Manus got acquired and if you haven't used Manus, it's actually really good. I think probably this year I'll start to talk a little bit more about some of these kind of startup super agent type companies. I think maybe when they first came out a year ago, I don't think that most of them were quite legitimate enough for most enterprise companies here in the US to putting to be putting resources behind. However, I think that's changed in the last quarter. I mean you've seen big companies such as Microsoft partner with genspark, which is in kind of like an equivalent or a very close competitor to Manus. And here we saw Manus get acquired by Meta. So some of the biggest players in the US tech scene are investing heavily or acquiring some of these general use super agents and you know, live stream audience. If you want to see a little bit more coverage on the super agents, I don't know, leave a, leave a comment, just say super agents or Spotify listeners. I always have a random number in my head and I say hey, but now people say this, I'm going to do a show on the next, you know, on this in the next couple of weeks, so let me know if you want to see that. But regardless, pretty big move. And let's be honest, Meta has been hurting in the AI game. So back in what, April of 25, they came out with their last model, Llama 4, which did not really do well, right? According to the benchmarks and just adoption, it was seemingly extremely low. So Meta's been in the in the kitchen cooking since then. They haven't come out with any major AI model releases. They've had a lot of great audio, visual AI releases, their SAM models, but nothing on the large language model side or the agent side until this acquisition of mana. So this may be a big part of their future plans moving forward. Next piece of AI news. Is OpenAI really interested in acquiring Pinterest? Well, according to the information, the answer is yes. That's because a new report from the Information says that OpenAI has interest or has reported interest in acquiring Pinterest, the not startup, I don't know what you call anymore a social media pinning site, right? It's really in a category of its own with 600 million users and it could become an acquisition target this year for OpenAI. So OpenAI's reported interest centers on Pinterest, vast image data and established advertising business, both seen as valuable for expanding OpenAI's online shopping and ads capabilities. Integration of OpenAI's advanced search and conversational AI with Pinterest visual platform could create new ways for users to interact, shop and in discover products. Also, Pinterest digital scrapbooking features may complement OpenAI's new image and video generation tools, potentially leading to innovative new user experiences. So despite its niche appeal, Pinterest faces slower growth in monthly average users and average revenue per user in the US and Canada, trailing larger rivals like Meta, Alphabet and Amazon. Also, you have to look at Pinterest's new ad suite called Pinterest Performance plus, launched last year, that helped advertisers achieve a 24% higher conversion lift compared to traditional campaigns. So outbound clicks to Advertisers rose by 40% in the third quarter for Pinterest, highlighting their strong user purchase intent in value to retail brands. So at first when I saw this headline I was like, that makes no sense. But then when I thought about it, I'm like, wait, this actually makes complete sense because what are the areas where OpenAI is really trying to compete in? Well, they're trying to be kind of The Everything AI app for everyone. Right. Nearing now a billion weekly active users. So, you know, you might think of a company like, I don't know, Microsoft or Anthropic in an acquisition like this might not make much sense, but we've seen OpenAI be obviously heavy players on the consumer side, where maybe Google, Anthropic, aws, et cetera, much more focused on revenue from enterprise companies. A vast majority of OpenAI's revenue comes from individuals. It comes from the consumer side, which is why we've seen a lot of the recent developments over the past year from OpenAI. Really focused on personal productivity, focus on things like shopping and, you know, bringing agentic commerce to ChatGPT and to your phone. Right. We saw a deep research mode launch specifically for shopping right around the holidays. So when you think of it like that and just OpenAI's play on the visual intelligence side, it makes sense. And this is something that I think could help them maybe instantly compete with the likes of Meta and Google, at least when it comes to ads in the shopping side. Because right now I do think Google is obviously so far ahead of everyone else and then the meta platforms are right there, but it's just really them. It's Google at 1a, meta at 1b, and then I don't know, I don't even know if there's anyone in the two position. Definitely no one else in the, in the first tier. So if this acquisition does come to fruition, I do think that OpenAI could immediately kind of skip five years, five to 10 years that a company would normally have to go through, right? Whether trying to compete on the search on the SEM PPC social advertising side, I mean, they can literally leapfrog five to 10 years of what would normally be needed and instantly be a top four or three company when it comes to advertising. So when you think of it like that, not the most surprising potential acquisition for OpenAI. All right, our next piece of AI news. LM arena has released its end of year 2025 ranking. So if you don't know LM arena, it is where you can go in, type a prompt, you get two blind outputs, and users choose which one is better. And then those are categorized across different categories like, you know, text, text to image, web development, all these different things. And pretty big news, right? So LM arena did just release their rankings of the top models across all categories. And I'll say this, Google Gemini 3 Pro cleaned up. It ranked as the top overall frontier model of 2025 and became the first AI model to reach an Elo score of 1500. So the Elo score essentially it's kind of like points right when users vote for a certain model. And even back when we in the like 1200s right at the time we were like my gosh, it's crazy that these models are getting scores of 1200, but here we are with models getting 1500. So in general text and reasoning tasks, Gemini Pro Gemini 3 Pro finished first with Gemini 3 Flash in second place. So there's fast cheap model in second place, pretty impressive. And then there's other non Google challengers such as Grok for one thinking Claude Opus for five and others. Gemini 3 Pro also ranked number one in multimodal vision tasks and coding performance as well. Gemini 3 Pro took the top spot over Claude Opus 4.5. Then search and grounded generation an important one, Gemini 3 Pro with grounding led in that area with GPT5.2 search right behind. So LM arena data also showed strong progress in generative video where surprise, surprise, Google's VO3.1 model led the text to video ranking. So probably one of the only areas that main ones I would say that Google didn't completely sweep, which is surprising, is the text to image model where GPT image 1.5, a late entry in December, came in and stole the top crown. Even amid all of the Nano Banana Pro buzz, that one, I'd say it's hotly contested, right? Because it really depends on what you're looking at. Text to image can mean like a dozen very different things. And at first when I saw GPT 1.5 image get the crown over Google Gemini, I'm like over there. Nano Banana Pro, I'm like, there's no way that can't be right. And then after looking at some of the categories, doing some tests myself, I'm like, okay, yeah, I can see it, but it is super, super close. So I wouldn't be surprised if Nano Banana Pro actually comes out with an iterative update and under the under the radar update, something like that to leapfrog because it is very close. But regardless, I think it's important to know that these are blindly tested. And hey, if it comes to 2025, it looks like Google won. And the people with across millions of votes seem to agree. All right, next, and this starts today. CES 2026, the consumer electronics show is already shaping up to be a more crowded and ambitious show this year than in recent years. And one of the reasons is, well, you have Nvidia CEO Jensen Huang keynoting the CES show starting any hour now. Maybe by the time you're listening to this podcast, Jensen Wong will be delivering his keynote address. So not expecting any huge news. That usually happens, at least on the Nvidia side, that usually happens at Nvidia's GTC conference. So. However, we might see some new Nvidia updates when it comes to their foundation models, Blackwell Ultra, some of their chips. So look at markets potentially moving just after, you know, Jensen Wong's keynote address. But why am I talking about CES when it hasn't even happened yet as one of the top AI news stories of the week? Well, that's because over the next week you're going to see a ton of both very interesting and absolutely ridiculous headlines because, well, that's what happens at ces. And especially over the past year or two, companies have been sticking AI in places it doesn't belong, like toasters and mops and shoes, right? So expect to see a lot of the same silly AI infused product announcements, mainly just geared at grabbing headlines. I have gotten a lot of information on embargo with, you know, products that are being released today. Nothing really great if I'm being honest. But if you are someone that enjoys technology, if you like gadgets and if you like AI. Oh wait, that's me. It is still going to be really fun to see some of the new announcements that come out of the CES show this week. And I do think one of the most dominant trends this year will obviously be AI being stuffed into every category from laptops to phones to appliances and cars. Also, I think smart glasses are expected to fly flood the show floor with both Meta and a bunch of startups that will be releasing some AI powered smart glasses. So it looks like Meta may finally have a little bit of competition and some of those are going to be unveiled today and the rest of the week at CES Livestream. Audience, what do you want to see? AI. What should AI be shoved into and what should it not be shoved into? Right, like, do I need to have a, a water bottle with AI that talks to me and tells me I'm drinking too much water? It's funny. My, my wife and cousin, they're always laughing that they don't drink enough water and maybe they need an app to, you know, help encourage them to drink more water? I don't know. If I had AI into this big water bottle, probably be like, you're drinking far too much water. All right, our next piece of AI news, a disturbing and disgusting one. So Elon Musk's AI Chatbot Grok is under intense scrutiny, both from millions of online users, governments and everyone after users online generated explicit images of children, including a 14 year old actress from the Stranger Things series. So the controversy is especially significant because Grok is authorized for official government use in the US and the chatbot has been running rogue, generating explicit images of miners on the platform whenever any user asks it to. So Grok, the actual chatbot, not the company X AI, but the automated AI powered chatbot, publicly admitted on X formerly Twitter, I just call it Twitter that there were quote unquote isolated cases where users created AI images of minors in minimal clothing, despite the company's terms of service prohibiting such content. So far, a couple governments, even though it's only been a couple of hours since this disgusting controversy, has erupted and gone viral in a bad way. But government officials in both India and France have launched official investigations into acts, with France referring the matter to an agency responsible for enforcing the EU's Digital Services act, which carries the potential for large fines. And then India's IT minister has given Xai, the company behind Grok, 72 hours to detail how it will stop the spread of illegal and explicit content involving minors. So the incident comes just months after the US government signed an 18 month contract to use Grok for federal business, despite warnings from over 30 consumer advocacy groups about the platform's safety and ideological bias. So XAI and X have not yet responded to requests for comment. But like I said, Grok the Chatbot posted that it is, quote unquote, urgently fixing its safeguard to prevent child sexual abuse materials. 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, 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 Gen AI. This is absolutely disgusting. I don't think I need to offer much commentary on this, but if you're like, wait, what the heck is going on? How does this work? Right? And this isn't new, right? This feature and functionality. So essentially, if anyone has shared an image online, I don't know when this functionality became available, but it's been at least three, three to five months, right? If anyone uploads a picture. So if I uploaded a picture to myself on Twitter, you could go in and just act, tag or mention Grok and say, you know, hey Ch, change Jordan's hair color, make it less gray, make it green, you know, make Jordan wear a tuxedo, right? And Grok will go in and edit the image and keep everything else the same. And to Grok's credit, and this is probably in a bad way, it actually has very good image and video models. Well, needless to say, people on the Internet who are gross and should probably be banned from using the Internet and the Grok chatbot have turned to using this for illicit and explicit purposes. And in the case of minors, and there's also obviously women are being targeted with this as well. Just, you know, this seen this in other reports as well, where women just sharing normal photos of themselves and then obviously Internet trolls doing bad things to those photos. So hopefully this is one of those instances where I hope the US government steps up and does something, although I can almost guarantee the US government will not. But it is at least somewhat positive and reassuring to see other countries come in and put the pressure on Grok and XI immediate XAI immediately. So hopefully this is a disturbing trend that will stop very quickly. All right, our next piece of AI news. An actual technically startup product that is generating more than a billion dollars is being developed by itself. So a major shift in AI assisted software development has been revealed as Boris Czerny, the creator of Claude Code at Anthropic, confirmed that in the past 30 days, every line of his project contributions were written by the AI itself. Yeah, let me say that the creator of Claude Code, which itself has generated more than a billion dollars of revenue in 2025, the creator of Claude Code said that all of his contributions in the past 30 days to Claude Code have been written by Claude Code. Crazy, right? So Attorney, the lead developer of Claude Code, stated on Twitter that 100% of his code contributions were generated by Claude Code itself, with none of it written by himself manually. So over the 30 day period, the AI submitted almost 260 PRs or pull requests, added more than 40,000 lines of codes, and made 497 commits on the project. Czerny described his role on Claude code as shifting from hands on coding to acting as an architect in verifier highlighting a new workflow where the main bottleneck is decided by is deciding just what to build and verifying correctness, not actually writing code. So Claude Code, which uses Anthropic's Claude Obus 4.5 model, operates as an autonomous coding agent directly in the developer's terminal. Also, Claude or Anthropic did just release a version of Claude code that works in the browser. So if you've never really used it or heard of it, the tool can plan features, debug issues, navigate file systems, and handle Git workflows, acting as a full coding partner rather than a simple assistant. So Cherni said he runs multiple instances of Claude code in parallel, dramatically increasing development speed and throughput. So, I don't know y', all, should we be covering Claude code a little bit more? When it first came out, I was like, probably not where we're at now, especially once Anthropic added Claude code. Well, a couple things now or previously, you couldn't use your normal, right? If you have the 20amonth Claude subscription that didn't give you access to Claude code originally. So I'm like, not really something to worry about. I think our niche, our sweet spot here is the tens of millions of business users who use large language models on the front end, not necessarily backend developers, although that line as well is starting to blur. However, I think once Anthropic release Claude code to the front end so you can go to Claude AI with your normal subscription and you can start using Claude code. But also now there's a lot of features and functionalities that you can use Claude code that is not just for development, not just for software engineering. Right. You can use Claude code now just to perform knowledge work tasks on your computer, right? Very much like a, you know, super agent of Manus or Gen Spark or chat GPT's agent mode. Right. Claude Code is actually capable of running some of those standard day to day tasks. So yeah, another not having to, you know, not trying to make you guys shout out random words in today's episode. But let me know, just say Claude code if you guys are interested or if you think we should be covering it a little bit more here on the show. All right, and our last piece of big AI News for the week Pen. A pen. Yes, a pen. OpenAI is reportedly working on an AI powered pen. Okay, so according to the information, OpenAI is actively developing a new consumer AI device and recent leaks suggest that one of the possible form factors is a smart pen designed for lightweight, always on available help from AI. So the pen is reportedly one of three hardware concepts currently being explored as OpenAI works with different manufacturing partners in their first foray into AI hardware. So the pen would likely combine handwriting capture and voice recording with ChatGPT powered assistants, similar to past smart Pension, but with deeper AI analysis and context awareness. So multiple reports indicate the device would feature always on listening, allowing users to ask questions, summarize conversations or process notes without pulling out a phone. OpenAI CEO Sam Altman has described the intended experience as feeling like a quote unquote cabin by a lake. Not talking about the pen, but just talking about AI hardware, signaling a calm, low distraction alternative to to smartphones. It's also been called kind of a third device, right? So not a phone, not a computer, but a third hardware device powered by AI that users would presumably be able to easily carry with them. So the internal code name for the project is reportedly Gumdrop, though no launch timeline or final design has been confirmed. So other concepts said to be under consideration include a wearable PIN and a smart speaker, suggesting OpenAI AI is experimenting rather than committing to a single device yet. So let's not confirm that the three different hardwares, right, it's not saying that one is a pen, one is a pin and one is a smart speaker. So so far we have confirmation, according to reports, that one of the three at least is a pen, like the pen that you write with. And the other ones not confirmed yet, but said that they could be a pin, like something that you, you know, might clip on your shirt and wear around all day, and maybe a smart speaker. So these efforts come as standalone AI hardware has struggled mightily in the market and with products not doing so great, such as the Humane AI pin that went belly up and rabbit R1 that didn't exactly go belly up. That had to do a pretty hard pivot in order to keep going. So that is the big AI news stories for the week. But as always, there's a lot of other recent smaller pieces of news. Rumors, rants, leaks, what's Next, what's new? So here we go. As we end most Monday AI news shows, we're giving you the quick bullet point version of what's new and what's new, what's next? All Right, here we go. So leaks show that Mistral is text testing workflows and shared connectors in beta Anthropic purchase close to a million Broadcon TPU V7 units for infrastructure expansion. According to SEMI analysis, OpenAI has reorganized teams to build audio based hardware AI products. Related to the story we just talked about there, XAI has launched Grok Business and enterprise, although I would guess most companies aren't going to be using that. Speaking of xai, they've also acquired a third Memphis data hub, nearly doubling their AI computing capacity. LM arena, like I said, announced their top frontier models notion is rolling out early access to its AI first workspace Fal Fal released its own version of the Flux 2 image generator called Flux Point 2 Turbo. Shaq. Yeah, AI news is so everywhere now Shaq is even in it. Shaq did a vibe coding ad in partnership with Replit, which he's an investor in. OpenAI announced an incubator type program called Grove in applications, opened up for the next cohort of technical founders. Epic AI released its benchmarks for GPT5.2 Pro after a little bit of a delay, which shows it has a massive lead in frontier math. So one area that Gemini did not claim, at least According to Epic AI benchmarks, ZipU AI, the maker of the GLM47 model. Hopefully I said that right, but Zai, formerly or commonly known as launched in IPO to list as Hong Kong's first LLM developer on their market. A new study showed that 20% of videos shown to new users on YouTube are straight up AI slop. Yeah, there was one of those trending this weekend. It got like a hundred million views. Like a random video of a monkey. Terrible, right? Did anyone else see that? Just me. Quinn's new open source layered LORA enables Photoshop grade layered editing of AI images. Very cool. And I do think this will be the future of AI images. But pretty cool to see from Quen, right? So you don't get a flat jpeg. You can use an AI image generator to generate images with layers that are all editable. Pretty cool. And then last but not least. Oh, a $41 billion piece of news sneak snuck in at the end, but technically not new news. But it was reported that Softbank did complete its total $41 billion investment in OpenAI according to the information. Kind of. According to their previous agreement, they were contractually obligated to fulfill about the second half of that 40 plus billion dollar commitment to OpenAI if OpenAI did complete its organizational restructure in 2025, which it did. So a ton of AI news this week. I hope this was helpful. If so, please if you're listening to the podcast, tell someone about this, please subscribe and follow the show on Spotify and Apple Podcasts. And that's it for today. But like I said, big week coming up. Not only are we going to be launching the Everyday AI Inner Circle community, but we also have some other fun announcements coming on the show this week. Some shows that we have technically been planning for months. So if one of your big goals this year is to just become better at AI, to make your company AI native, whatever it may be, you are not going to want to miss a single episode this January. I kid you not. If you are, you're going being behind because we've been planning for this month for months. Not going to want to miss it. 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.