OpenAI Declares Code Red

Today on the AI Daily Brief, we have a code red from OpenAI. They are gearing up for battle as the narrative shifts over to Gemini, and that means potentially way more goodies for all of us. Before that, in the headlines, Apple's AI lead is finally leaving. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. All right, friends, quick announcements before we dive in. First of all, thank you to today's sponsors, KPMG, Blitzy, Rovo, and Robots and Pencils. To get an ad-free version of the show, go to patreon.com slash AIDailyBrief, or you can subscribe on Apple Podcasts. And if you are interested in sponsoring the show, head on over to AIDailyBrief.ai to learn all about sponsor opportunities. You can also send us a note at sponsors at AIDailyBrief.ai. Last quick one before we dive in, we've got a new product we're testing at Superintelligent that I am super excited about that I want to give you first access to. If our agent readiness audits are designed for companies that are moving from zero to one and want help with their overall AI planning, we're running into tons of companies that are somewhere stuck in the middle. They've got some pilots, they've got some initiatives, but they're not getting the full value that they actually want. For them, we've designed a new program called the Plateau Breaker program that in just a handful of weeks is going to diagnose exactly why your AI pilots and initiatives are stuck. And we're going to build a full blueprint an action plan to get unstuck based on our experience across hundreds of companies and successful AI strategies. If you're interested to learn more, shoot us a note at contact at bsuper.ai with plateau in the subject line. Again, that's contact at bsuper.ai with plateau in the subject line. With that though, let's dive into the headlines. Welcome back to the AI Daily Brief Headlines Edition, all the daily AI news you need in around five minutes. As you'll see from the main story, it seems like December 2nd is already providing a ton of evidence for everything we talked about on December 1st, which was, of course, our predictions for what the big stories in December would be. We kick off the headlines today with the announcement that Apple's AI lead is finally out, marking the end of an underwhelming era. Apple announced that John Gianandrea will be stepping down as Senior VP of Machine Learning and AI Strategy before exiting the company entirely next spring. Gianandria joined Apple from Google in 2018 and helmed AI efforts at the company until earlier this year. In March, Bloomberg's Apple Insider Mark Gurman wrote that CEO Tim Cook had lost confidence in Gianandria's ability to execute on product development. At that time, leadership of the Siri product was transferred to Vision Pro creator Mike Rockwell, who would report to SVP of software engineering Craig Federighi cutting Gianandria out of the loop on Apple's most urgent AI product. In other words, Gianandria's exit seemed inevitable for almost the entire year, but it is a sign of how lumbering, frankly, and slow-moving Apple has become that it took nearly nine months after Tim Cook apparently lost confidence for this ship to be made when this is such an important product category. Now, Apple apparently won't directly replace the role, instead opting to split the artificial intelligence team under multiple new leaders. Staff will report to SVP of software engineering, Craig Federighi, COO, Savick Khan, and SVP of services, Eddie Q. Eddie Q, you might remember, was one of the folks who was most aggressively pushing acquisitions as a way for Apple to get back into the race, I think if I remember correctly leading the charge around the perplexity talks that never ultimately materialized. One new addition is Amar Sabramania, who is joining Apple as VP of AI reporting directly to Federighi. In that role, he will lead foundation model development, machine learning research, and AI safety and evaluation, which had been Gianandria's core responsibilities over the recent months. Sabramania most recently served as corporate VP for AI at Microsoft and was previously head of engineering for Gemini at Google. Now, Swix underscored one big point, posting, this is burying the lead. He joined Microsoft AI four months ago, which either means Apple made him a great offer or he really didn't like working at Microsoft. Now, Bloomberg's German noted that Apple still has the odds stacked against them. They're currently working towards a spring release for the new AI Siri after missing the intended release date this year. Siri will now be powered by Google's models rather than models developed in-house at Apple. And of course, the AI team has seen a huge amount of attrition, More than a dozen members of the model team have left, including lead scientist Ro Ming Pang, who joined Meta over the summer. I would say that people's general take on this is good riddance, but not sure if this is enough. Mark Gurman writes, Strange hire for a number of reasons, but it's hard to argue the Apple job is a bad one. Anything is an improvement at this part, so the bar is as low as it comes. Easy layup on the resume. Mac Daily News writes, An unmitigated failure. Steve Jobs would have bounced this bozo on the day ChatGPT was released in November 2022. Gianandria finally and ridiculously slowly and gently being shown the door is better years late than never, we guess. Listen, ultimately, we asked in our December preview episode if we were going to see signs of life at Apple, and this certainly seems to be a necessary step for any of that to proceed. Next up, a little model news. As we were recording yesterday's episode, we got the drop of Runway 4.5, as well as a new model from DeepSeek, And we also got Kling's Video 01, which they're pitching as the world's first unified multimodal video model. The big deal here, and the thing that people are excited about, is basically nano-banana style editing, but in video. So the idea is that you can swap protagonists and elements, weather, styles, colors, etc. Which, as it did with photos, if it works well, opens up an entirely new set of use cases for generative video. I'm hoping to come back and do a broader episode about Kling 01 and Runway 4.5. First impressions outside of the hypesters are, I think, a little bit disappointed. The lack of native audio really hinders the general consumer use possibilities of these models, but they might be orienting towards their niche in professional production. OpenAI and their Code Red are the subject of our main episode but they also make the headlines after taking an ownership stake in Thrive Holdings Now this deal on first glance is reinvigorating the discussions of circular dealmaking in AI Thrive Holdings is a subsidiary of Thrive Capital who are one of OpenAI's major investors. The subsidiary was launched in April and functions as a private equity roll-up vehicle with an AI twist. At the moment, they're focused on acquiring professional services firms like accountants and IT support, then adding AI tools to boost efficiency. This is the big PE play of the moment right now, with lots of folks trying some version of this strategy. OpenAI plans to embed engineering, research, and product teams within Thrive's portfolio companies to help accelerate the strategy. Jared Kushner, the CEO of Thrive Capital, said, We are excited to extend our partnership with OpenAI to embed their frontier models, products, and services into sectors we believe have tremendous potential to benefit from technological innovation and adoption. He added, For decades, technology has transformed the world's largest industries from the outside in. We believe the AI paradigm will be different in that some of the most profound transformations will now occur from the inside out. We view the businesses that we own and operate as the right reward system to build, test, and improve industry specific products and models. Now, interestingly, while the circularity of the deal is sort of an obvious component, there are actually kind of a lot of folks who see the business logic here. Milkroad AI writes, OpenAI will work with Thrive Holdings to accelerate how businesses adopt AI starting with accounting and IT services firms. So you've got OpenAI's investors funding a holding company and then OpenAI itself taking equity in that same holding company. In other words, and this is back to NLW's language, this is less OpenAI investing in the fund that invests in OpenAI and more OpenAI in one of its investors collaborating on a subsidiary under one of those investors that is an operating business that takes advantage of OpenAI's technologies. Going back to Milk Road, they continue, it's a closed ecosystem where capital and technology flow together. Thrive Capital funds the acquisitions, Thrive Holdings owns the operating companies, and OpenAI provides the AI integration and strategic guidance while capturing upside as equity owner. The strategy is simple. Don't just sell AI. Own the businesses that become valuable because of it. David Hendrickson writes, By controlling Thrive Holdings, it gives them an environment to test its models. Selling AI to big enterprises is slow because they're risk-averse and have old systems. Through this investment, Thrive Holdings buys these service companies, and OpenAI embeds its engineers directly into them. They can experiment with workflow-heavy rules-driven tasks without needing to convince an external client to take the risk. And I think that as someone who is deep inside exactly this sort of transformation right now, there is absolutely a clear logic to this. Having a testbed with fewer barriers to try radical and complete AI transformation is extremely valuable. and also it's not clear ultimately if the better business model is trying to transform the old companies through enterprise sales or just owning the new actors who take advantage of the new technology better to become the dominant companies of their age. Overall, I think the response to this has been more positive than I might've imagined, which is good for OpenAI because they need momentum right now as we will see in the main episode. In fact, let's close the headlines here and jump on over into OpenAI's Code Red. Hello, friends. If you've been enjoying what we've been discussing on the show, you'll want to check out another podcast that I've had the privilege to host, which is called You Can With AI from KPMG. Season one was designed to be a set of real stories from real leaders making AI work in their organizations. And now season two is coming and we're back with even bigger conversations. This show is entirely focused on what it's like to actually drive AI change inside your enterprise, and as case studies, expert panels, and a lot more practical goodness that I hope will be extremely valuable for you as the listener. Search You Can with AI on Apple, Spotify, or YouTube and subscribe today. This episode is brought to you by Blitzy, the enterprise autonomous software development platform with infinite code context. Blitzy uses thousands of specialized AI agents that think for hours to understand enterprise-scale code bases with millions of lines of code. Enterprise engineering leaders start every development sprint with the Blitzy platform, bringing in their development requirements. The Blitzy platform provides a plan, then generates and precompiles code for each task. Blitzy delivers 80% plus of the development work autonomously while providing a guide for the final 20% of human development work required to complete the sprint. Public companies are achieving a 5x engineering velocity increase when incorporating Blitzy as their pre-IDE development tool, pairing it with their coding pilot of choice to bring an AI-native SDLC into their org. Visit Blitzy.com and press Get a Demo to learn how Blitzy transforms your SDLC from AI-assisted to AI-native. Meet Rovo, your AI-powered teammate. Rovo unleashes the potential of your team with AI-powered search, chat, and agents, or build your own agent with Studio. Rovo is powered by your organization's knowledge and lives on Atlassian's trusted and secure platform, so it's always working in the context of your work. Connect Rovo to your favorite SaaS app so no knowledge gets left behind. Rovo runs on the Teamwork Graph, Atlassian's intelligence layer that unifies data across all of your apps and delivers personalized AI insights from day one. Rovo is already built into Jira, Confluence, and Jira Service Management Standard, Premium, and Enterprise subscriptions. Know the feeling when AI turns from tool to teammate? If you Rovo, you know. Discover Rovo, your new AI teammate powered by Atlassian. Get started at ROV, as in victory, O.com. Small, nimble teams beat bloated consulting every time. Robots & Pencils partners with organizations on intelligent, cloud-native systems powered by AI. They cover human needs design AI solutions and cut through complexity to deliver meaningful impact without the layers of bureaucracy As an AWS partner Robots Pencils combines the reach of a large firm with the focus of a trusted partner With teams across the US Canada Europe and Latin America, clients gain local expertise and global scale. As AI evolves, they ensure you keep peace with change. And that means faster results, measurable outcomes, and a partnership built to last. The right partner makes progress inevitable. Partner with Robots and Pencils at robotsandpencils.com slash AIDailyBrief. Welcome back to the AIDailyBrief. Today we have an exciting one, some very quick follow-up to our five big stories to watch in AI episode from just yesterday. Now, going back to November, it was very clear that the big story was everything surrounding Google and its AI strategy. So that includes Gemini 3, Nano Banana Pro, the rise in performance of their TPUs and what it means for NVIDIA, and specifically how all of these things change the competitive landscape for AI. Never before in the three years, the exactly three years since ChatGPT launched, have we seen such a narrative inflection shift from OpenAI over to Google. More and more, you were starting to see the language of inevitability around Google, which for some was all-inclusive and for others, maybe more specific to things like multimodal AI. But no matter what, OpenAI was facing very serious competition, not just with the tool itself, but also with public perception. Now add to this, of course, that Anthropic dropped Opus 4.5, and that model has done nothing but get increasingly rave reviews the longer it's been out, All of which comes together to last night, when actually after I went to sleep, the information came out with an exclusive report around a warm memo from Sam Altman to the team at OpenAI. Basically three years to the week after the launch of ChatGPT led Sundar Pichai and Google to declare their own code red, now the roles have been reversed. And Sam Altman on Monday of this week told employees at OpenAI that he was declaring a code red to focus all of their resources on improving their core asset, which is ChatGPT. So what does it mean to actually improve ChatGPT? Well, it sounds like there are a couple pieces. Altman apparently mentioned more personalization for the chatbot, including letting ChatGPT's users have more customization around the way that it interacts with them. Altman also said that they wanted to improve model behavior. So as the information put it, people prefer the AI models that power ChatGPT more than models from competitors, including in public rankings such as LM Arena to other priorities including boosting chat GPT's speed and reliability as well as minimizing over-refusals. Over-refusals in this case refer to when chat GPT refuses to answer something that is actually a benign non-harmful question. I guarantee that you have had that experience and it is just enormously frustrating when you're not asking it for anything complicated or scary or challenging and yet it still refuses to answer. It is one of the most frustrating experiences you can have with a chatbot So it makes sense if they're trying to improve the experience of ChatGPT to cut down on those sort of blockers. Now, as part of this announcement, it sounds like we're not just getting a renewed and reinvigorated commitment to ChatGPT, but an actual pause on some other initiatives. Allman, it sounds like, explicitly said that to focus on ChatGPT, they were going to delay other initiatives. The one that everyone is talking about is advertising. Now, I've talked a bit about OpenAI's communication around advertising. frankly in a frustrated way. I think that when Sora 2, the model, was launched alongside Sora, the application, it felt pretty inevitable to a lot of people that OpenAI was going to move into ads at some point. It just seems like the only way to support a full-on video-based social network. Also, there are clear opportunities for ad integration into the core ChatGPT experience. And yet OpenAI, specifically Sam Altman, has continuously been super circumspect about it, saying that no decisions have been made rather than doing what I think he should have done, which is basically say ads are a great way to keep things free. Ads don't have to suck. We care most about user experience and we're not going to let ads screw that up. And while we have made no final decisions, there is likely an ad integrated future for many of these products, which you will get to see a mile in advance because we respect you as the user and are going to treat you like adults as we talk to you about our plans here. In other words, I think that OpenAI or Altman might be slightly overestimating people's unwillingness to deal with ads and underestimating their frustration at feeling tricked or deceived. In any case, to the extent that there are ad things going on, those plans are now on pause. The information also writes that Altman said the code red surge to improve ChatGPT meant that the company would, quote, delay progress on other products such as AI agents, specifically those aiming to automate tasks related to shopping and health, as well as developments on their Pulse product, which is the thing that generates those personalized reports for ChatGPT users to read each morning. Additionally, there was some new model talk. Now, you'll remember that in the predictions for December episode, I said that I thought that the most likely area where we would see a new OpenAI model this month would be around image generation, not even just to regain momentum, but to keep feature parity and people operating within the OpenAI ecosystem. It does sound like this is a priority, but it's not clear from the reports that we got how far along their next generation image generation model is. Writes the information, Altman said that other key priorities covered by the Code Red include the image-generating AI that allows ChatGPT users to create anything from interior design mock-ups to turning real-life photos into animated ones. No mention on how far along their next version of ImageGen actually is, but one big nugget that basically everyone is seizing onto is this one. Altman said that the company is planning to ship a new reasoning model next week that they claim is ahead of Google Gemini 3 in OpenAI internal evaluations Now the information adds that Altman said that the company also had more work to do on improving the chat GPT experience. And while I think it's right that those experience upgrades make a huge difference when it comes to things like how long people are spending in the app and how much they're switching to other models, when it comes to just narrative momentum, there is certainly nothing bigger that OpenAI can do than drop a reasoning model that actually feels ahead of the competition. Now, there's certainly chatter that those models exist. In a podcast appearance, OpenAI research leader Mark Chen said, we have models internally that perform at the level of Gemini 3, and we're pretty confident that we will release them soon, and we can release successor models that are even better. ChatGPT leader Nick Turley dropped a Twitter thread around the time the information story broke, seemingly to try to set some of the framework for where the state of the competition is right now. He wrote, Today, ChatGPT is the number one assistant worldwide with around 70% of assistant usage. New products are launching every week, which is great. It pushes us to move faster and keep raising the bar for what an AI assistant can do. In a clear nod to Google, he continues, Search is one of the biggest areas of opportunity. ChatGPT now accounts for roughly 10% of search activity, and it's growing quickly. As more people ask ChatGPT not just for answers but to take action on their behalf, we see lots of room to expand. He concludes, Our focus now is to keep making ChatGPT more capable, continue growing, and expand access around the world, while making it feel even more intuitive and personal. So let's talk about how people receive this. Overall, the vibe is excitement. People are fired up. Part of that is just because competition is fun to watch, but part of it is because it benefits us. However, one thing that people noticed that was kind of conspicuously absent, at least from the reporting, was that there was no mention of coding models. Swix called this out in his post about Code Red, and there were many posts like this one from Bill Berkey who wrote, Dear Sam Altman, I read in the Wall Street Journal this morning that OpenAI's Altman declares Code Red to improve chat GPT as Google threatens AI lead. Make codecs work, dot to dot. Sincerely, Vibe Coders. Now, two thoughts about this. First of all, it may be that Altman and OpenAI are making a determination that the flank that they're feeling most threatened on when it comes to Google Gemini is general consumer usage that doesn't really touch the vibe coding space. Remember, we saw a bunch of charts recently around how Google Gemini is catching up in terms of monthly downloads and is now ahead in terms of time per session. And so it may be that leaving coding off is intentional because they don't see that as a major front in their battle with Google, at least not right now. The information points out that broad perception of the competition between OpenAI and Google is going to be key to OpenAI's ability to raise the additional hundreds of billions of dollars they're going to need over the next few years. And so as much as coding might matter to a lot of enfranchised users, maybe they're determining that it's not the thing that drives those top line numbers that the casual investor, for example, is going to read. The other thing to note, though, which is always an important caveat, is that we are seeing this secondhand. We are reading a report based on what Sam dropped on Slack that was read by the reporters, not by us. If the reporter who saw it missed a mention of coding or just decided it wasn't important, we could have the wrong perception just because they are dictating what we're seeing. It's also possible that the tone of the Slack message wasn't a super comprehensive list of everything that's going to be worked on versus everything that's going to be deprioritized, instead being something that's a little bit more off the cuff and much more about the emotional and team rallying cry than the specifics, in which case it would be easy to read too much into any one thing that's included or not. Now, some think that Sam might know that whenever he writes something like this, it's going to get leaked. And so was perhaps using it to generate exactly the type of response that he's getting from me covering it on this show. Beth Jezos writes, I bet the whole Code Red story was a psyop and OpenAI is about to drop a massively better model. That post was actually a repost of AI leaker Jimmy Apple's account that had a picture of Altman with a Santa hat on and the message hype, hype, hype, hype, hype, hype, hype, hype, hype, hype, hype hype hype. While I'm recording before markets open on Tuesday, Knockout Trader did point out that SoftBank had a bit of a week overnight session after the Code Red news broke and said, we'll be interesting to see if this pressure will find its way towards US names being linked to OpenAI in any way. For example, will we see ripple effects in Oracle stock tomorrow? Overall, though, mostly people are just super excited. Jeffrey Snover captured my feeling exactly when he wrote, three cheers for competition. OpenAI made Gemini get better, and now it's the other way and we all benefit. Look, man, I was tearing down to December where I didn't think we were going to get all that much in terms of new models. And now it seems like that is very much not the case. And we could be playing with a new advanced reasoning model from OpenAI in the very near future. It is definitely the case that the company that crystallized this field for so many now finds itself on the back foot. If you go to the prediction market sites, the betting around questions like which company has the best AI model by the end of 2025 is distinctly in Google's favor right now. And the gap has been increasing, not decreasing. Although it is worth noting that after the news broke in one market on Polymarket, Google dropped from 92 to 88%. And OpenAI jumped from 0.5% to 7.6% in just a few hours. All in all, I think we have a major competition on our hands and I'm not counting anyone out. So friends, there you have it. Code Red has been declared. and now we get to see what happens next. That's going to do it for today's AI Daily Brief. Appreciate you listening or watching as always. And until next time, peace.