Could Gemini 3.0 and Claude 4.5 Be Coming Soon?

0:00Today on the AI Daily Brief, the AI rumor mill really wants us to get Gemini 3 or Claude 4.5. Is it possible that we might see that in the next few weeks? Before that in the headlines, one prominent investor thinks that AI could shrink public companies by a fifth. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI.

0:24All right, friends, quick announcements before we dive in. First of all, thank you to today's sponsors, KPMG, Robots and Pencils, Notion, and Agency.org. To get an ad-free version of the show, go to patreon.com slash AI Daily Brief. And a quick note for those of you who have applied for the account management role that I mentioned on the show, I think maybe last Wednesday, appreciate you all sending your videos of your AI workflows. It's a great way to get to know you. Much, much better, I will say, than a resume and a cover letter. We're going to be trying to dig into all of that and get back to folks, at least preliminarily, next week, so look out for that.

0:53With all that said, let's get into these comments from the CIO of Allianz. Welcome back to the AI Daily Brief Headlines Edition, all the daily AI news you need in around five minutes. We kick off today with a big pronouncement from the CIO of Allianz Global Investors, who believes that AI could usher in a period of digital Darwinism. Speaking at the Bloomberg Investment Management Summit in London, Virginie Mezanouf warned that this decade is going to be very, very brutal for public companies that fail to adapt. She said, AI is both a threat and a huge opportunity. It's very hard to make a forecast, but 15 to 20 percent of companies listed today could not be here in five years time.

1:32Now, it is worth noting that that isn't actually wildly above the norm for the mortality rate of publicly listed companies, but you get the sense certainly that she is talking about major well-known companies rather than the normal attrition of smaller forgettable firms. And frankly, her comments are more useful as a check-in and where we are in the AI adoption narrative. I.e., although some headlines are still blaring that AI has questionable return on investment based on even more questionable studies, professional investors like this at the very top of their industry now see AI adoption as a key marker of a firm's ability to survive the decade.

2:04To take just one example, Bloomberg also published a piece this week called The AI Threat to Europe's Most Valuable Software Company. The article is about SAP and its challenges to adapt to changing technology paradigms. The gist of it is that while it has finally made a successful transition to cloud. After years of struggling with that, the company's growth in the cloud area is expected to diminish significantly after 2027, meaning that it's going to have to transition to new types of products like AI products to maintain its current trajectory. And it's pretty clear that the CIO is again not talking about the normal base churn of low to mid-cap companies, but about these sort of more dramatic changes.

2:43Although speaking of SAP, OpenAI is partnering with SAP to build Sovereign AI for Germany. The initiative will see OpenAI deploy their technology on SAP servers that are physically located in Germany. OpenAI wrote, the collaboration will enable millions of public sector employees to use AI safely and responsibly while meeting strict data sovereignty, security, and legal standards. The infrastructure will use Microsoft's Azure hardware, but will be set up to fit the requirements of the German government. OpenAI also said that the project would serve as a cornerstone for, quote, developing customized public sector applications and integrating AI agents directly into existing workflows to automate and improve processes such as records management and administrative data analysis.

3:21Now, speaking of questions of sovereign AI and global AI leadership, global AI governance was up for debate at this week's meeting of the United Nations. The session opened on Wednesday with Secretary General Antonio Guterres saying, The question is not whether AI will influence international peace and security, but how we will shape its influence used responsibly. AI can strengthen prevention and protection, anticipating food insecurity and displacement, supporting demining, helping identify potential outbreaks of violence and so much more. But without guardrails, it can also be weaponized. Now, heading into this meeting, a group of about 200 public figures, including former world leaders and Nobel Prize laureates, had called for binding international red lines, proposing an international agreement defining what AI should never be allowed to do.

4:02To me, the response to this letter signified just how much this group is frankly talking to themselves right now. Now, part of that is AI specific. I think that basically all of these open letters and calls for pauses have done nothing but actually diminish the voice of those people who are calling for better guardrails around AI. But of course, this is also caught up in broader shifts in international politics and U.S. leadership as well. Speaking of which, White House Office of Science and Technology Director Michael Kratios dismissed the notion that the U.N. should play any role in governing AI out of hand.

4:32He said, in incredibly clear terms, we totally reject all efforts by international bodies to assert centralized control and global governance of AI. We believe that the responsible diffusion of AI will help pave the way to a flourishing future, one of increased productivity, empowered individuals, and revolutions in scientific advancement. The path to this world is not found in bureaucratic management, but in the freedom and duty of citizens, the prudence and cooperation of statesmen, and the independence and sovereignty of nations. We believe broad over-regulation incentivizes centralization, stifles innovation, and increases the danger that these tools will be used for tyranny and conquest.

5:05Ideological fixations on social equity, climate catastrophism, and so-called existential risk are dangers to progress and obstacles to responsibly harnessing this technology as an extension of human ingenuity and capacities. Could not be clear that, like basically everything else when it comes to the UN and global governance, the current US administration is completely uninterested in any sort of actual cooperation with those groups. Now, as we've talked about in the absence of government-level regulation, the private sector has been exploring its own ways to deal with some of the issues that come with AI.

5:34On that note, Cloudflare has introduced a new addition to robots.txt to limit AI crawlers. Of course, as you've heard, over recent months, Cloudflare have been raising the alarm about what they perceive as the problem of AI driving down traffic to websites. They've made some big proposals like a new fee market to charge AI crawlers to scrape websites. They've also taken some aggressive high-profile moves like blocking crawlers without customer approval and calling out perplexity specifically as a bad actor. Now, their latest proposal goes back to basics with a new addition to robots.txt, which is a file on every website that tells crawlers whether or not they're allowed to scrape data.

6:06They're calling their change the Cloudflare content signals policy. The default format for robots.txt doesn't differentiate between different types of crawlers. Cloudflare then has now proposed three categories, search, AI inputs, and AI training, which can each be toggled independently. In explaining the logic behind the traffic, Cloudflare noted that they expect bot traffic on the internet to exceed human traffic by 2029. And by 2031, they expect that bot traffic will surpass the current level of human traffic. The rise is entirely due to AI search engines who tap website data but don't drive any human traffic or produce ad revenue.

6:38Cloudflare writes, the de facto default of the internet permitted this. The norm has been that your data would be ingested, but then you, the creator of that content, would get something in return, either referral traffic that you could monetize, or at a minimum, some sort of attribution that cited you as the author. They continued writing, that world has changed. That scraped content is now sometimes used to economically compete against the original creator. It's left many with an impossible choice. Do you lock down access to your content and data or accept the reality of fewer referrals and minimal attribution?

7:06If the only recourse is the former, the open transmission of ideas on the web is harmed, and newer entrants to the AI ecosystem are put at an unfair disadvantage for their efforts to train new models. Now, many outlets recognize that this change is fairly squarely aimed at taking on Google. While the changing nature of internet traffic is broadly due to the rise of AI search, Google's AI overviews have of course been singled out as a major culprit. Earlier this month, the publisher of Rolling Stone, among other magazines, sued Google over AI overviews, with the core of their claim being that Google provided no way for websites to agree to allow search indexing without also agreeing to their data being used in AI overviews.

7:40Exactly the sort of differentiation that Cloudflare is trying to provide with this new toggle. Now, on the one hand, this is a real debate, but on the other, not everyone is super happy with Cloudflare appointing themselves as the guardian of this particular issue. Jeff Jarvis, for example, wrote, Who died and made Cloudflare cop of AI? Now, one more very cool one from Google, regardless of your stance on AI overviews. Google has plugged their massive trove of publicly available data into the AI world using MCP. Back in 2018, Google launched a project called the Data Commons with the goal of collating all the publicly available data in the world.

8:13The archive includes government surveys, local administrative data, and statistics from global bodies like the UN. However, the data set was sprawling, making it difficult to query unless you knew exactly what you were looking for. Google figured that this was a perfect problem to solve with AI, writing, the release marks a major milestone in making all of Data Commons vast and interconnected public datasets instantly accessible and actionable for AI developers, data scientists, and organizations worldwide. This is the type of very cool use case that standard infrastructure can enable, and so I will be excited to see what developers do with this new trove of information that they now have easier access to.

8:48For now, though, that is going to do it for today's AI Daily Brief Headlines edition. Next up, the main episode. What if AI wasn't just a buzzword, but a business imperative? On You Can With AI, we take you inside the boardrooms and strategy sessions of the world's most forward-thinking enterprises. Hosted by me, Nathaniel Whittemore, and powered by KPMG, this seven-part series delivers real-world insights from leaders who are scaling AI with purpose. From aligning culture and leadership to building trust, data readiness, and deploying AI agents. Whether you're a C-suite executive, strategist, or innovator, this podcast is your front row seat to the future of enterprise AI.

9:24So go check it out at www.kpmg.us slash AI podcasts, or search You Can With AI on Spotify, Apple Podcasts, or wherever you get your podcasts. AI changes fast. You need a partner built for the long game. Robots and pencils work side by side with organizations to turn AI ambition into real human impact. As an AWS certified partner, they modernize infrastructure, design cloud native systems, and apply AI to create business value. And their partnerships don't end at launch. As AI changes, Robots and Pencils stays by your side, so you keep pace. The difference is close partnership that builds value and compounds over time.

10:03Plus, with delivery centers across the US, Canada, Europe, and Latin America, clients get local expertise and global scale. For AI that delivers progress, not promises, visit robotsandpencils.com slash AI Daily Brief. Chatbots are great, but they can only take you so far. I've recently been testing Notion's new AI agents, and they are a very different type of experience. These are agents that actually complete entire workflows for you in your style. And best of all, they work in a channel that you already know and love because they are purpose-built Notion super users. Notion's new AI agents completely expands the range of what Notion can do.

10:39It can now build documents from your entire company's knowledge base, organize scattered information into organized reports, basically do tasks that used to take days, and get them complete in minutes. These agents don't just help with work, they finish it. Getting started with building on Notion is easier than ever. Notion agents are now your very own super user to help you onboard in minutes. Your AI teammates are ready to work. Try Notion AI for free at the link in our show notes. Shape the future of enterprise AI with Agency. A-G-N-T-C-Y. Now an open source Linux foundation project, Agency is leading the way in establishing trusted identity and access management for the Internet of Agents.

11:15the collaboration layer that ensures AI agents can securely discover, connect, and work across any framework. With agency, your organization gains open, standardized tools and seamless integration, including robust identity management to be able to identify, authenticate, and interact across any platform. Empowering you to deploy multi-agent systems with confidence, join industry leaders like Cisco, Dell Technologies, Google Cloud, Oracle, Red Hat, and 75-plus supporting companies to set the standard for secure, scalable AI infrastructure. Is your enterprise ready for the future of agentic AI?

11:46Visit agency.org to explore use cases now. That's agntcy.org. Welcome back to the AI Daily Brief. Some days on this show, we get quite heady. Big new ideas, research papers, technical innovations that could change the world. Other days, we're super practical. I've got an episode coming up about how to avoid the scourge of work slop. And then On other days, we are just digging into the rumor mill, checking in on the vibes, and understanding the state of competition among the big labs. Today is one of those days, and this is something that's been brewing for a little while. If you have been watching the sentiment closely, it feels like there was a big shift around Anthropic just over the last month or so.

12:29Anthropic bullposter account SuperDario tweeted, Word is Anthropic team at Offsite this week trying to regroup and get back on feet after inference stumbles and launch delays. dedicated to sticking the landing. Down but not out, still bullish. Chubby writes, it's strange how the vibe is turning against Claude and increasingly positive towards OpenAI and ChatGPT. The former bubble darling, Anthropic, is increasingly losing favor. A couple of days ago, Elon Musk himself waded in saying winning was never in the set of possible outcomes for Anthropic, leading to posts like this one from Mark Kretschmann who wrote, it's over for Anthropic and I'm not going to miss them.

13:05So what is going on? Why is Anthropic in this low ebb right now when it comes to reputation? There are a couple of reasons. Anthropic has had, by any estimation, an absolutely unbelievable year. The company grew revenue from about a billion to five billion, and for most of the year really cemented and confirmed its dominance as the coding model of choice. That dominance, however, looked a little bit rocky towards the end of the summer. People were complaining left

13:38Now, Anthropic has said over and over again that they were not intentionally throttling their models and that they would never do that. But they did come out with a sort of mea culpa explanation of what was going on. Summing up the problem, Unite AI writes, For the past six weeks, Claude users have been losing their minds. Starting in early August, complaints started flooding Reddit, X, and developer forums. The issues were all over the place. Code that used to work perfectly was suddenly broken. Claude would claim it made changes to files when it didn't. Random tire Chinese characters appeared in English responses.

14:07Instructions were being completely ignored. The same prompt was giving wildly different quality responses. And Claude code users said it felt lobotomized compared to before. The complaints got so bad that by late August, people were convinced Anthropic was secretly throttling Claude to save money. Conspiracy theories were everywhere. Maybe they were reducing quality during peak hours. Maybe they'd quietly swapped in a cheaper model. Maybe this was intentional degradation to manage server costs. Turns out that the problem, they say, was not any sort of intentional throttling, but actual bugs that, as they put it, intermittently degraded responses from Claude.

14:40Now, we went through this on a previous show, but for our purposes today, the important point is that there were some big performance issues that created an opening, and it's an opening that OpenAI seized. GPT-5 was very clearly, from the moment it was launched, aimed at winning back some of the market share around agentic coding that OpenAI had lost to Anthropic over the previous year. And while initially, the response to the model was largely drowned out by all the issues surrounding weird choices like the deprecation of old models, issues with the model selector, etc., people slowly started to come around largely because of GPT-5's coding capabilities and Codex more specifically.

15:17Slow developer Hayter on Twitter writes, GPT-5 Codex is the best product launch of Q4 2025. It follows instructions, sticks to the guidelines, doesn't overcomplicate, and produces optimized code. If you're into vibe coding, it might not be for you, but if you know what you want, it beats Cloud Code in every way. Chongyu points out that Codex has more GitHub stars than Cloud code, even though it was released a month and a half later. More generally, it's just clear that for the first time, Claude has some real competition when it comes to being the default code model, and that has gotten worse with the release of GPT-5 Codex.

15:48So a big part of the vibe issues are just simply the fact that Anthropik's model performance was degraded for a little while, right at the time when one of their fiercest competitors was releasing something that was, for the first time, actually competitive for that use case. Now, there also have been some other issues. The company's leadership has been standing on principle vis-a-vis Donald Trump. The Wall Street Journal recently reported that CEO Dario Amadei is, in their words, openly challenging President Trump's hands-off AI strategy. He explicitly did not come to that White House dinner for AI leaders, and that's put him and the company directly at odds with AI czar David Sachs, who has been taking potshots at Dario and Anthropic for months now.

16:23Still, while that creates political headwinds that are a pain in the butt for them to deal with, I don't think that's actually at the core of any of these issues of the vibe they are currently experiencing. I certainly think more at issue is self-inflicted wounds around some of the things that Dario has chosen to speak about in recent interviews. Here's, for example, what he said recently about open source. I don't think open source works the same way in AI that it has worked in other areas, primarily because with open source, you can see the source code of the model. Here, we can't see inside the model.

16:57It's often called open weights instead of open source to kind of distinguish that. But a lot of the benefits, which is that many people can work on it, that it's kind of additive, it doesn't quite work in the same way. So, you know, I've actually always seen it as a red herring. When I see a new model come out, I don't care whether it's open source or not. Like if we talk about DeepSeek, I don't think it mattered that DeepSeek is open source. I think I ask, is it a good model? Right. Is it better than us at, you know, the things that that's the only thing that I care about? It actually doesn't matter either way because ultimately you have to host it on the cloud.

17:34The people who host it on the cloud do inference. These are big models. They're hard to do inference on. And conversely, many of the things that you can do when you see the weights, we're increasingly offering on clouds where you can fine-tune the model. You can, you know, you know, we're even looking at ways to, you know, to to kind of, you know, investigate the activations of the model as part of like an interpretability interface. We did some little things around steering last time. So I think it's the wrong axis to think in terms of when I think about competition, I think about like which models are good at the tasks that we do.

18:11I think open source is actually a red herring. So that may just be Dario thinking from first principles, but it was not received in an open and thoughtful way. Instead, the much more common response was summed up by this one from Mark Kretschmann, who wrote, Dario Amadei is showing his true face, anti-competitive doomer with a love of regulation to control AI. For that reason, he hates open source AI. Anthropic has always been like that, but many of you are late to notice it. Now, even people that weren't that caustic, like Cyril Zaka from Hugging Face said, Rare miss from Dario. Open source AI has been one of the most important drivers of innovation and optimizations in the field.

18:46Quite disappointing take. There was a lot more at various ends of that spectrum, but wading into this pond at this moment when they were already dealing with some narrative issues was sort of just adding on to a fire that was already burning. Now the counter narrative to all of this comes from Lisan Algaib who writes, What's up with the whole anthropic doomer sentiment today? They will drop the best coding model once again and keep growing at 10x a year. And I do think that it's always worth asking when there is some weird vibe shift, how much it has to do with people just not having something else to talk about and asking the question, what have you done lately for me?

19:18And on that front, the rumor mill has been swirling for some time now that we were about to get a Claude 4.5 release. Now, if SuperDario is right, maybe that release has been delayed and that's part of the issue here. But there is certainly also a sentiment that once that happens, there are going to be a whole lot of people who are willing to forgive any issues they had in the past if the model actually is as good as people hope it will be. Another positive thing that Anthropic has going for them is the Keep Thinking brand campaign they launched last week, which is certainly the highest quality traditional sort of brand messaging that we've seen from any AI company so far.

19:49Former White House AI leader Dean Ball writes, I mean it as the absolute highest compliment when I say that of all the AI companies today, none remind me more of early 2000s Apple than Anthropic. Attached to that tweet was the core asset of that Keep Thinking campaign. And the other interesting note is that while the enfranchised chattering classes may be on a comparative low ebb with their relationship with Claude, the models themselves are potentially getting their biggest audience ever with the new deal recently announced by Microsoft. We had reported early as this started to leak, but this week Satya Nadella and Microsoft made it official, tweeting, Today we're expanding Microsoft 365 Copilot with the addition of Anthropics Claude models.

20:26Customers can now use both OpenAI and Claude starting in research in Copilot Studio and coming to more experiences soon. Our multimodal approach goes beyond choice. It's all about bringing the best AI firm across the industry to co-pilot, tuned for work, and tailored for every business. Ultimately, I tend to be in the LaSanne camp here. I think that we are in a period where Anthropic is shaking off the dust of degraded performance, which unfortunately came at a moment when OpenAI became a realistic competitor for coding applications for the first time in more than a year. But foundation model companies always look worse when one of their competitors has recently released a model and they haven't.

21:00All I think it's probably going to take to get people back on board is a next generation clawed model, assuming that it actually is better on the coding front than the tools we have access to today. Now, expanding this into a little bit of a roundup of where the labs are in general, we've heard about Anthropic and OpenAI. Recently, I also told the story of that town hall where Satya Nadella of Microsoft said that he was haunted at the prospect of Microsoft not surviving this AI era. I think the interesting thing about Microsoft right now is that they have clearly, firmly, and finally broken off their trajectory from their first big AI partner of this era in OpenAI.

21:32I don't think it's some big dramatic thing. I just think that they're taking a different path, and increasingly that path is coming into view. Writing about why Microsoft had passed on all of this Stargate build-out, Reuters wrote, Through late 2024, Microsoft, then OpenAI's exclusive cloud partner, was scouting new data center sites for OpenAI. As OpenAI's compute appetite ballooned, Microsoft began to re-evaluate all the properties it was leasing to find space for expansion. Any property that wasn't in prominent locations or didn't have a solid power contract locked in was put on the chopping block.

22:01Oracle and SoftBank stepped in to source sites with over 1 gigawatt requirements. Microsoft, meanwhile, kept expanding at the edges of existing regions, avoiding one gargantuan bet. They continue, Microsoft now favors clusters sized for long-term reuse with staged GPU refreshes, distributed across regions to support inference as much as training. Here's Microsoft's bet. The future of AI isn't another colossal supercomputer in one location. It's a fast distributed web of AI power serving billions of people globally. The winning AI platform will be the one that delivers a fast, seamless experience to users no matter where they are in the world.