Resources:
Join the discussion on LinkedIn: Got something to say? Let us know on LinkedIn and network with other AI leaders
Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineup
Connect with Jordan Wilson: LinkedIn Profile
Start Here Series in our Inner Circle Community: Join for free access
AI News That Demands Attention: Anthropic’s Compute Deal, OpenAI’s Model Upgrade, and Government Model Testing Set New Standards
AI is transforming at a pace that now requires sharper discernment among business owners and decision makers. The most recent developments—many under the radar—signal critical implications for enterprise reliability, regulatory oversight, and workforce adaptation. Here’s a comprehensive translation of the latest AI news into direct business value and actionable insights.
Anthropic Leverages SpaceX Compute: Addressing Reliability for the Enterprise AI Market
Anthropic, regarded as a chief competitor to OpenAI, has entered a landmark compute leasing agreement with Elon Musk’s XAI, utilizing the Colossus 1 supercomputer—reportedly housing 220,000 NVIDIA GPUs. This deal, potentially adding $3–4 billion in annual revenue to SpaceX (with $2.5 billion in cash profit), is not a mere transaction but a response to a central pain point in enterprise AI: reliability.
Anthropic has faced scrutiny for suboptimal uptime—reportedly in the 98% range, well below the “three nines” or “four nines” (99.9–99.99%) typically expected from critical enterprise software. This newly secured compute capacity aims to raise reliability standards, directly addressing the concerns of finance and other sectors seeking AI deployments resilient enough for operational continuity. The move exemplifies the new stakes in AI infrastructure: Those providing consistent service, not just innovation, will capture the enterprise wallet share.
AI Job Cuts: Cloudflare’s “AI-First” Restructuring Unpacks New Market Risks
Cloudflare’s recent 20% workforce reduction—impacting 1,100 positions—is notable not for its size but for strategic rationale. Described as the company’s largest job reduction ever, the initiative is part of an “AI-first” restructuring, aligning internal operations toward greater automation and agentic tool adoption.
Unlike prior cost-cutting announcements, this move coincided with lower-than-expected revenue guidance, leading to a 20% decline in Cloudflare stock. The incident counters Wall Street’s usual response to “efficient” AI-driven layoffs and signals to technology leaders and investors that operational timing and messaging around AI job adjustments must sync closely with company performance. This moment sets a precedent: job reductions attributed to AI will no longer automatically be rewarded unless profitability and revenue data support the strategy.
Apple’s AI Promises: $250M Settlement Highlights Importance of Delivery over Hype
Apple’s $250 million class action settlement, stemming from claims of overstated AI capabilities in its “Apple Intelligence” suite post-iPhone 16 launch, demonstrates the tangible business risks of overpromising. Advanced writing tools and a conversational Siri, heavily advertised, remained largely undelivered, drawing regulatory scrutiny and consumer backlash.
The included payment eligibility for customers who purchased certain phone models in 2024 and 2025 reinforces the necessity of honest marketing around AI features. Notably, the episode underscores the longer-term financial impact of missed AI investments—potentially trillions in market cap lost due to a slower pivot toward AI R&D. The lesson: competitive positioning in AI must be underpinned by transparent capabilities, not rebranding or unfounded claims.
OpenAI GPT-5.5 Instant Rolls Out: Enterprise-Grade Accuracy and Contextual Memory
The transition to GPT-5.5 Instant as the default OpenAI ChatGPT model, replacing 5.3 Instant, has meaningful technical and business implications. With a score of 81.2—an improvement over 65.4 on the AIM 2025 math test—GPT-5.5 delivers accuracy and speed leveled up for enterprise use. Most notably, enhanced context management capabilities allow integration with past chats, files, and Gmail to personalize responses, first to Plus and Pro users and soon to enterprise tiers.
Contextual memory and the ability to reference exact answer sources—along with tools for information management—open direct pathways for auditability, compliance tracking, and internal QA, especially in sensitive areas like finance and legal workflows. For developers, understanding the three-month migration window away from 5.3 is crucial for supporting ongoing integrations.
Microsoft’s Work Trend Index: Organizational Pace Limits AI’s Business Impact
A global survey of 20,000 AI-using knowledge workers, featured in Microsoft’s 2026 Work Trend Index, uncovers a persistent transformation bottleneck: only 19% of respondents work in environments with both strong organizational AI support and skilled individuals—the so-called “frontier group.” The majority remain constrained by lagging management practices, misaligned incentives, and outdated metrics, only 26% of users report that leadership is unified and clear on AI strategy.
The research pinpoints a structural hurdle: 67% of AI effectiveness is attributable to culture, manager support, and policy, versus just 32% from individual readiness. Organizations that act decisively to reorganize talent practices and management buy-in around AI adoption will reap outsized productivity dividends, leaving slow-movers vulnerable to competitive gaps regardless of individual upskilling efforts.
Federal Testing Becomes Prerequisite: Implications for AI Product Roadmaps
The formation of the Center for AI Standards and Innovation (CAISI) within the US Department of Commerce now sets new requirements for advanced AI models from key developers, including Microsoft, Google DeepMind, and XAI. Federal agencies will conduct pre- and post-deployment testing, focusing on security, risk, and real-world misuse, with government officials signaling increased oversight for releases such as Anthropic’s Mythos system.
This marks a 180-degree change in US government posture, moving from passive monitoring to hands-on intervention. Any company building on top of major AI platforms will need to factor centralized federal vetting—and potential delays—into product timelines. Further, the expanding agreement circle signals that AI product teams must proactively anticipate regulatory hurdles, especially in models with network or cyber implications.
Model Deployment as a Service: OpenAI and Anthropic Expand Into Consulting
Both OpenAI and Anthropic are now establishing or acquiring engineering and consulting organizations (“Deployco” for OpenAI, an unnamed deployment unit for Anthropic) to bridge the gap between model capability and business adoption. This goes beyond selling API access: it’s an admission that deploying, customizing, and integrating AI within unique enterprise workflows requires specialized labor and hands-on adaptation—mirroring Palantir’s embedded services model.
With billions raised to acquire existing engineering and consulting firms, the competitive axis is shifting from pure R&D speed to integrated delivery. For clients, this means broader, more tailored deployment options and, likely, industry consolidation as small- and mid-tier consultancies become acquisition targets or lose relevance if unable to rapidly upskill.
Conclusion
These developments, each precise and quantifiable, point to a new era for AI in business, characterized by merciless scrutiny of reliability, transparency between claims and capabilities, the necessity for cultural and managerial adaptation, and mandatory regulatory navigation. Strategy now demands a holistic approach: robust infrastructure, honest marketing, employee upskilling paired with management change, and proactive compliance, all delivered through specialized talent. Successful organizations will not be those that dabble in AI, but those that build repeatable infrastructure and practices qualified for enterprise and regulatory standards.
Topics Covered in This Episode:
- SpaceX Leases Colossus Supercomputer to Anthropic
- Cloudflare Restructures, Cuts 20% Jobs for AI
- Apple Settles $250M AI Feature Lawsuit
- OpenAI Releases GPT-5.5 Instant Model Update
- ChatGPT Personalized Context and Memory Features
- Microsoft 2026 Work Trend Index on AI Adoption
- White House Moves Toward Federal AI Model Testing
- OpenAI, Anthropic Launch AI Consulting Services
- OpenAI Voice Models, Real-Time Translation Rollout
- Google DeepMind Staff Unionize Against Military AI
- Anthropic Expands Claude to Office Suite
- Adobe Launches AI-Powered PDF Collaboration Tools
Episode Transcript
Jordan Wilson [00:00:18]:
You know what most companies pitching AI agents aren't actually doing? Deploying AI agents. Sage just did the opposite at Sage Future. They expanded their AWS collaboration to put real AI agents into real finance workflows built on Amazon Bedrock agents core and available in the AWS marketplace. So that means enterprise grade AI for your finance team without the IT headache. Check it out at sage dot com. The AI tools that we use every single day have completely changed, and hardly anyone noticed. That's the importance of keeping up with AI news, but let me lay all of this out. I mean, 900,000,000 people each week are likely using a new model inside of chat GBT and probably don't know.
Jordan Wilson [00:01:14]:
Claude could become more reliable for once, and it took a very unlikely third party to make that happen. And a new AI study had some pretty shocking findings about how bad most companies are at deploying AI in the workforce. Oh, and the federal government might soon have its fingerprint on every single new AI model that comes out in a kind of a 180 degree flip on its previous stance. If you miss any of those AI news stories, don't worry. We're gonna be recapping those and a whole lot more today on everyday AI. What's going on y'all? Welcome to everyday AI. My name is Jordan Wilson, and this thing's for you. It's your daily unedited livestream podcast and free daily newsletter helping everyday business leaders like you and me not just keep up with all these AI developments, but what they mean and how to use this information to grow your company and your career.
Jordan Wilson [00:02:06]:
So if that's what you're trying to do, me too. It starts here, but make sure you go to youreverydayai.com. Sign up for our free daily newsletter, and we're gonna be recapping all of these stories and a whole lot more. So, without further ado, let's get into it. Who is that third party that might be helping Anthropic keep up with the demands that they just can't quite serve? Well, it is OpenAI's chief rival now Elon Musk in, their company XAI. So here's what's new and why you need to know. So Elon Musk has leased SpaceX's largest and most valuable supercomputer to Anthropic despite previously calling the company evil and accusing it of hating Western civilization, making this a pretty sharp and newsworthy reversal. So the deal centers on XAI's Colossus 1, a massive data center in Memphis containing about 220,000 NVIDIA GPUs, and analysts estimate it will generate 3 to $4,000,000,000 in annual revenue for SpaceX with more than $2,500,000,000 in cash profit.
Jordan Wilson [00:03:19]:
So the agreement between, XAI, which is now under the SpaceX, Well, might be now called SpaceXAI. Elon keeps changing the names, but the agreement comes just weeks before SpaceX is expected to begin its IPO roadshow after filing a confidential s one, in April targeting evaluation between 1.5 and $2,000,000,000,000 according to analysts. So the timing also matters for Anthropic. That's because the Anthropic lease does give SpaceX a marquee AI customer, and, hopefully, what Anthropic is hoping that their service will become more reliable before they start to get closer to their, plans to go public. Colossal, Colossus one was originally built in 2024 to train Musk's AI assistant, Grok, but Grok generates right now under $1,000,000,000 in annualized revenue while Anthropic is on track to exceed maybe closer to $30,000,000,000 according to some reports, creating excess compute that SpaceX was motivated to monetize. So by leasing directly to Anthropic, SpaceX captures the high margins that AI labs usually pay to the hyperscalers like AWS, Microsoft Azure, or Google Cloud whose compute margins exceed 30% according to most analysts. So, here's the interesting part. Right? Anthropic, their uptime is is pretty bad.
Jordan Wilson [00:04:53]:
Right? So across most industries, you aim for three nines or four nines. Right? So 99.9% uptime, 99.99% uptime. Right? That's kind of your gold standard for most enterprise software. So, as a lot of enterprises now are looking at Infropix claw to say, okay. Could this be our AI check out of choice? Yeah. Their uptime on some of these different services is in the 90 eights. So not even close, to where it should be for any reliable service, let alone something that's as critical as an AI, operating system, essentially. So, obviously, Anthropic and OpenAI have kind of been, you know, racing, especially in 2026.
Jordan Wilson [00:05:37]:
You know, I think now they're kind of seen as rivals, whereas before, maybe they weren't seen as much as rivals, you know, aside from people in the AI space every single day like myself. But for the most part, it really intensified in 2026. They started exchanging all these jabs, but one thing that Anthropic cannot compete on, at least right now, is uptime. OpenAI is extremely reliable, and Anthropic is not. So it is pretty interesting here that Elon Musk did call Anthropic evil not too long ago, and now he's essentially giving them all of not all the compute they need, but a lot of the compute that they need to, you know, actually challenge OpenAI. And, obviously, Elon Musk has been suing OpenAI. So it's almost like, you know, Elon calls Anthropic evil. He clearly doesn't like OpenAI, so he's, you know, teaming up with a company he called evil to go against a company that he probably likes even less.
Jordan Wilson [00:06:35]:
So pretty interesting here in terms of, you know, unlikely duos that most people probably weren't expecting. Alright. Our next piece of AI news, CloudFlare cutting 20% of its global workforce, which is about 1,100 jobs, marking the company's first mass layoff in its largest job reduction to date. So the layoffs were announced alongside cloud, Cloudflare's earnings reports, making that move newsworthy because it signals a major strategic shift rather than a short term response to financial trouble. So, obviously, company executives say the cuts are part of an AI first restructuring aimed at redesigning how work gets done as automation and so called agentic AI tools take on more across the business. So Cloudflare emphasized that the job reductions are not primarily about cost cutting or fixing performance problems, but instead are about reorganizing teams to move faster with AI driven operations. Investors, though, this one's interesting, reacted sharply with Cloudflare shares falling more than 20% after the company issued second quarter revenue guidance that came in slightly below Wall Street expectations. So that one is interesting because, normally, when big a big companies, especially AI companies that are public, you know, come out with news that they're cutting jobs because of AI, generally, their stock goes up.
Jordan Wilson [00:08:10]:
But this one is primarily, you know, with their stock going down, I think happens because the company said that their second quarter revenue came in slightly below expectations. So my thought is if Cloudflare's revenue came in above expectations and they cut it 20% of their workforce because of AI, their stock would have shot up because that does seem like the, kind of the blueprint that most companies have been following. So according to CloudFlare, the restructuring will affect roles across regions and departments, highlighting how AI adoption is beginning to reshape white collar tech jobs, not just manufacturing or support work. So if you don't know Cloudflare, they kind of run the entire Internet, which is why this one is probably important to anyone. Even if you've never heard of CloudFlare, there's a good chance that your website, your company's business runs on CloudFlare in some shape or form. Maybe if your company's website even even isn't even, you know, hosted or routed through Cloudflare's services, there's a good chance that a crucial piece of software, that you use on the day to day basis runs through Cloudflare. So it does have its thumbprint on literally the entire Internet and how everything operates, AI companies and, non AI companies as well. So it will be, worth noting to see how this rolls out because this is the first, kind of big AI company that announced AI job cuts, and their stock went down.
Jordan Wilson [00:09:43]:
So could this be a new trend, moving forward? Might this, you know, keep companies from announcing these large AI job cuts, especially public companies, when they see that Cloudflare's stock actually went down after this announcement? I would say probably not. My my thought is this is actually just going to be a key signal moving forward that when big companies announce they're cutting thousands of jobs because of AI, they have to do it more strategically around their earnings. And I do think that will actually be the big takeaway from this because most other companies when they announce, you know, these big AI job cuts, their stock is usually already going up or their earnings are, you know, meeting or exceeding expectations, and that's usually when they couple in these, you know, AI, job cuts. So, you know, who knows? Maybe CloudFlare wasn't able to sustain this much longer. But in their, you know, announcement, they essentially said, you know, hey. AI can can do the job way faster now, so we don't need all of these, you know, roles that we had previously have. But, the messaging, the timing, especially for public companies, I think CloudFlare is going to serve as one of those small use cases for other companies because I've been saying this since 2023. This is the playbook, right, for the most part.
Jordan Wilson [00:10:55]:
CloudFlare is one of the the few exceptions here. But you announce big AI job cuts. You show that revenue is going up, and, well, your stock will go up accordingly because one thing Wall Street loves, they loves, Wall Street loves efficiency. They hate employees, and they love profit going up. So, usually, you can pair those three things together, in these announcements. CloudFlare, the exception here. So alright. Let's keep moving our next piece of AI news.
Jordan Wilson [00:11:30]:
Apple has reached a settlement for promising AI that it couldn't deliver as part of its Apple intelligence. So Apple has agreed to a $250,000,000 class action lawsuit settlement after being accused of misleading customers about the capabilities of their Apple intelligence. So the settlement matters because it centers on how major tech companies market their internal artificial intelligence at a time when AI claims are shaping consumer buying decisions and regulatory scrutiny is now increasing. So the lawsuit claimed that Apple overstated AI features such as advanced writing tools and a more conversational Siri, which were heavily heavily promoted before in Apple the after the iPhone 16 launch, in September 2024, but were not fully delivered. Yeah. So if you go back to Apple's worldwide developer conference, WWDC twenty twenty four, everything was about, you know, Apple intelligence, and I scoffed at it, obviously, the fact that Apple was bold enough to try to rebrand artificial intelligence as Apple intelligence, but, I mean, I knew, and I think most people knew that Apple was so far behind and that all of these promises were not likely to come to fruition. Right? Go back. Listen to some of the shows around that time.
Jordan Wilson [00:13:00]:
I obviously called it out at the time, and, you know, I think Apple has probably ended up costing itself, multiple trillions of dollars in market cap over the long run by not properly, investing and prioritizing in AI over the past couple of years. We'll see. I think maybe their future, angle could actually work out best where they're just kind of admitting that they can't do it, and then they're paying, you know, competitors like Google to just use their AI instead. But here's, how this lawsuit is going to play out this class action settlement. So US customers who bought specific iPhone 15 and iPhone 16 models between June 2024 and March 2025 may qualify for payments ranging from $25 to $95 once a judge approves the amounts. So eligible devices listed in the agreement include the iPhone 15 pro, the iPhone 15 pro max, and the full iPhone 16 lineup, including the pro and pro max models. As part of this settlement, though, Apple did not admit wrongdoing and said in a statement to The New York Times that it settled the case to stay focused on building products and services for users. So the case combines several class action lawsuits and that argued Apple violated consumer protection laws by promoting enhanced Siri features and AI tools that were not ready or remain unavailable today.
Jordan Wilson [00:14:26]:
So Apple's delayed AI powered Siri has been a major industry talking point with the updated version now expected to fully roll out sometime around WWDC twenty twenty six, roughly two years later than early marketing suggested. So yeah. I mean, we'll see what Apple does with AI now that Tim Cook is out of the kind of day to day, CEO role and focus now, as chairman. But this, at least to me, comes as no surprise. Right? Apple having to settle a class action lawsuit because they overpromise on some basic AI features that they couldn't deliver. Right? And I do think, ultimately, this will be maybe studied, you know, in business school, you know, ten to twenty years or, you know, maybe by our AI tutors, in ten to twenty years, about how not to market features, right, especially when you just don't have you don't have it. Right? Apple didn't have really anything, and they were promoting all of these features that clearly did not work or they could not deliver on, which is interesting because Apple had the capabilities. Back if you go back to 2019, 2020, right, when the industry, for the most part, knew that this is where things were headed.
Jordan Wilson [00:15:49]:
Right? It wasn't the chat GPT moment, of November 2022 that signaled where the future of technology was headed. I think that most companies, especially bigger companies, you know, knew right after the, kind of introduction or the realization of the transformer, from the attention is all you need paper. I think most companies understood that the future was gonna be around AI. Right? And, obviously, Apple had the talent. Right? They had great machine learning teams already, up to that point. I don't know if they were too slow, if they were too proud. You know, one one kind of common story that we saw over the years was Apple just refusing to pay talent. Right? You know, it got to the point, I would say, especially in 2022, 2023, 2024, where you had these bidding wars.
Jordan Wilson [00:16:36]:
Right? All of a sudden, you know, the smartest AI leaders, the smartest, machine learning engineers were starting to demand million dollar salaries, and it seemed like Apple was just like, no. We're not gonna play. Right? All the stories, you know, meta, you know, nabbing Apple researchers, OpenAI nabbing Apple researchers, Google, everyone pulling the smartest AI people away from Apple, but you didn't see any stories about people going to Apple. And a lot of the the the reasoning and the rationale behind it was, well, Apple just wasn't gonna play the game. They weren't gonna pay anyone a million dollar salary, a $5,000,000 salary. Right? Some of these companies pay these people tens of millions of dollars over a multiple year contract. And Apple kind of from the outside was like, no. We're Apple.
Jordan Wilson [00:17:19]:
We don't do that. Or, you know, we're not gonna invest in AI. People you know, companies are still gonna use, you know, Apple products regardless. So, again, we'll see with new leadership, in play what happens here. Obviously, the surge of local AI models has probably helped Apple a lot, sell a lot more, you know, Mac minis and, you know, Mac Studios. You know, a lot of people flocking to buy those to run open models like, you know, OpenClaw and just to run models on their local machines. So maybe Apple has been a, kind of side beneficiary of the local model surge. But when it comes to actually delivering the capabilities, that consumers both want, and business users, Apple failed to follow through on their promises, and now they're having to dish out a quarter billion dollars.
Jordan Wilson [00:18:10]:
But if I'm being honest, you know, I'm I'm one of those users that's gonna get, you know, $90 or whatever because, you know, my whole family, we bought iPhones around that time. It wasn't just because of Apple intelligence. It was more of like, okay. Need need new iPhones, and we've been on iPhones for, you know, ten years. So if I'm being honest, I think Apple got off lightly here with only a $250,000,000 settlement, considering how heavy they promoted. Right? They literally had commercials showing Apple intelligence features that did not exist, and they actually aired those commercials. Right? And they weren't even close to existing. So I actually think Apple kinda got off light here.
Jordan Wilson [00:18:48]:
Alright. Our next AI news story, kind of a big one, but, technically, a small release that's gonna impact, like, I don't know, almost a billion people. That's because OpenAI has released a new model called GPT five five instance, which replaced GPT five three instant as the default chat GPT model, marking a major update focused on accuracy, speed, and real world usefulness. So the company says GPT 5.5 instant reduces hallucinations in sensitive fields like law, medicine, and finance while keeping the low response times that made the previous instant model popular. So GPT 5.5 scored 81.2 on the AIM 2025 math test, a sharp improvement over GPT three instance score of 65.4, signaling stronger performance in advanced problem solving tasks. So a key focus of the new release is context management, and this one kind of flew under the radar if I'm being honest. As GBT 5.5 instant now can search can use search tools to reference past chats, files, and even Gmail to deliver more personalized responses. So, this personalized context feature is rolling out first to plus and pro users on the web with mobile support coming soon and broader access plan for free, go business and enterprise users in the coming weeks.
Jordan Wilson [00:20:22]:
ChattGPT will now show memory sources across all models, allowing users to see where answers came from, delete outdated information, or correct mistakes while keeping the those sources hidden when chats are shared. So if you're on the developer side, GPT 5.5 will be available, via the chat latest naming with GPT five three remaining available for paid users for only, three more months. I mean, here's why this is a big deal aside from, well, it's just a better, more powerful, more capable model. Kind of a naming a little confusing here. Right? Because even before GPT 5.5 came out, for paid users, they had GPT 5.4, which was a very capable model. Right? I loved 5.4, and I think a lot of power users really loved it because I think with 5.4, especially 5.4 pro, that's kind of the the the time when OpenAI's most powerful model, and it still holds true across benchmarks today with 5.5 pro. That pro model can't be touched, at least not by Google, not by Anthropic. No one is getting close to the pro tier.
Jordan Wilson [00:21:36]:
Right? But there was, I think, a pretty big gap because even when you had GPT five four, the default model, the default chat model for users that didn't change anything was still 5.3 instant. So when we had 5.5, the the default instant model that most people were using was GPT five three instant. So it's technically, like, two small generational steps behind. Right? So this is why it's actually a pretty big deal, Not just because this is the default u, the default model that nearly a billion people are using, and it's that much better, but all of these new memory features that are rolling out as well that I think are actually gonna make a pretty big deal. Alright. We have more, but before we get into it, a quick break for a word from our partners. Adobe just introduced an entirely new way to create, bringing the power and precision of its creative suite into one conversational experience. Meet Firefly AI assistant, now live in the Adobe Firefly app, the all in one creative AI studio.
Jordan Wilson [00:22:39]:
Powered by Adobe's creative agent, Firefly AI assistant lets you start with your vision, just describe what you want, and shape the outcome as it takes form with the assistant. The assistant orchestrates multi step workflows, drawing on 60 plus pro grade tools across Adobe Creative Cloud apps, including Photoshop, Illustrator Premier, Lightroom Express, and more to help bring your ideas to life. You can also get started with creative skills, a growing library of pre built workflows for common creative tasks like batch editing photos, creating mood boards, portrait retouching, and creating social variations. Every step the assistant takes is visible, so you can refine, redirect, or take over at any time. You stay in the driver's seat as the creative director. Adobe Firefly AI assistant now in public beta. See it today @firefly.adobe.com. Alright.
Jordan Wilson [00:23:36]:
Our next AI news story, one of my favorite AI studies that I don't think gets enough shine and coverage. So I'm gonna talk about it today. So a new story, sorry, a new study from Microsoft's 2026 work trend index finds a growing gap between how fast employees are adopting AI and how slowly most organizations are changing to support it. So according to Microsoft's new study, and this is the work trend index annual report. They surveyed 20,000 AI using knowledge workers across 10 countries, and one of their biggest findings was only that 19% are in what they call the frontier group where strong individual AI skills and strong organizational support reinforce each other. Right? So that's kind of the gold standard when you have AI, you know, kind of AI native employees and you have strong organizational structure around them. But the remaining, workforce, this study found, is spread across weaker positions, including 50% in it in the emergent middle, 10% in the blocked agency where skilled AI users are actually held back by their companies, and 16% who are stalled, and 5% where companies are ready, but employees are actually not. So Microsoft calls this the transformation paradox, meaning employees are ready to change how work gets done, but company systems like incentives, metrics, and management practices keep reinforcing old ways of working.
Jordan Wilson [00:25:16]:
Organizational factors matter far more than individual talent, the study found, with culture, manager support, and talent practices accounting for 67% of reported AI impact compared with 32% from individual mindset and behavior. So, yeah, even when you have employees that have the right mindset and behavior, which is a pretty big, gap, it doesn't really matter too much because you do need the culture, the manager support, and the talent practices, to support those individual leaders. Leadership alignment is also a major weakness reported as only 26% of AI users say their leadership is clearly aligned on AI, and leaders are twice as likely as employees to believe that reinvention is rewarded, revealing a large perception gap. Microsoft found that the advanced frontier professionals who make up about 16% of AI users are far more productive, more deliberate about when to use AI, and far more likely to work for managers who actively model AI use and set quality standards. So, yeah, I think the big takeaway here is that 16%, right or no. I'm sorry. 19% in the frontier group. And that's when you have, essentially, individual employees who are bought in.
Jordan Wilson [00:26:40]:
They have been able to take advantage of the latest that AI, has to offer, and then companies that provide the proper support. So the fact that you only have 19% of that, I in the year 2026, the middle of 2026 is pretty telling. It is telling how, you you know, AI, I think, is no longer a capabilities problem in terms of, you know, individuals understanding the capabilities of AI because you still have the individuals who are ahead, but it's more of just getting the company aligned and onboard. Right? These companies that are slow moving ships and just wrapped in multiple layers of of bureaucracy, they're not ultimately going to come out of this AI transformation on the right side. It is gonna be those whether they're large enterprises or smaller businesses that from the organizational level understand this big ship of AI is not like anything that we've really ever seen, and you can't apply the same, you know, one year pilot, the same quarterly road maps that you applied to the cloud or mobile. So great, study from Microsoft. If you missed it in our newsletter, last week, we will link it again this week. Alright.
Jordan Wilson [00:27:55]:
Our next AI news story, the Trump White House may have their fingerprints on all future powerful AI models that come out of The US. So the White House is moving toward a federal AI model vetting, process. So this is pretty big because the US government is moving closer to a formal oversight of those powerful AI systems with federal agencies now positioning themselves to test advanced models before they reach the public. So here is the new group that you should probably pay attention to. It's a mouthful. So it is the Center for AI Standards and Innovation or the CAISI. I'm wondering if we could just call them Cassie. I don't know.
Jordan Wilson [00:28:40]:
But that's part of the US Department of Commerce's National Institute of Standards and Technology. So they've signed, new agreements with Google DeepMind, Microsoft, and Elon Musk's XAI to evaluate their AI models, before and after deployment. So according to the CAISI, the agency will conduct pre deployment testing in targeted research to assess frontier AI capabilities, focusing on security risk, autonomous behavior, and potential real world misuse. So these agreements expand a program first launched under the Biden administration when CAISI then called just the US Artificial Intelligence Safety Institute signed major deals with Anthropic and OpenAI nearly two years ago. So, obviously, we've heard what's happening with Anthropic, in the Department of Defense, but they did already have those preexisting, agreements with Anthropic and OpenAI. So the new deals with, you know, Microsoft, Google DeepMind, and XAI are added, to those companies that already had the preexisting deals. So Microsoft said in a blog post that government led testing is essential to building trust in advanced AI, arguing that as AI systems become more capable, the rigor of safeguards and testing must increase as well. So the announcement comes just ahead of a Bloomberg report that the White House is preparing an executive order to create a centralized federal vetting system for new AI models, including Anthropic's recently unveiled Mythos system, right, that they've been really hyping up but have not yet released to the public.
Jordan Wilson [00:30:20]:
So Bloomberg also reported that Anthropic disclosed Mythos was highly effective at identifying network vulnerabilities, raising concerns inside the administration about potential global cyber cybersecurity risks if such systems are misused misused. Alright. Our last big story of the week, Open and OpenAI and Anthropic are getting into the AI services firms. Right? So according to reports, OpenAI and Anthropic are in talks to both acquire and build engineering and consulting companies that help businesses deploy AI, signaling a pretty big shift in the AI race from just these companies building the models themselves. So OpenAI's new joint venture is called the Deployment Company or Deployco, and they are right now in advanced talks on three major acquisitions, and they're raising about $4,000,000,000 from 19 big institutional investors, including t b g TPG, Bain Capital, and Brookfield Asset Management according to sources. So, yeah, they're raising a ton of money from some of the biggest backers in the game. Anthropic is reportedly making a similar move, raising about $1,500,000,000 from investors including Blackstone, Hellman and Freeman, and Goldman Sachs as it builds its own deployment focused ventures. So, the goal of both of these companies is to bring in hundreds of engineers and consultants who can customize AI models for specific company data systems and workflows, which remains a major hurdle to widespread enterprise AI adoption.
Jordan Wilson [00:32:06]:
So this marks a new competitive front for AI companies as success as success increasingly depends not just on how powerful your models are and how good your AI features are, but more on the labor intensive work required to actually integrate them into everyday business operations. So reports show that most of the capital raised by the two companies will likely be used to buy smaller engineering services and consulting firms, potentially consolidating a fragmented market. So, this strategy mirrors Palantir's long standing approach of embedding their engineers inside client organizations to adopt software on the ground, a model now being adopted more widely across the AI industry. To me, this is no surprise. Why? Well, because I've been saying, I don't know, since, like, 2024, that the consulting the big consulting groups were too slow. Right? A lot of when the chat g b t moment first happened, you had some of the bigger consulting firms kind of poo pooing on AI. They're like, nope. We're not gonna use it.
Jordan Wilson [00:33:12]:
You know, we are, you know, essentially smarter. We don't need AI. Right? We have all of our own systems. And then eventually, you know, by for the most part, by 2024, most of the big consulting firms kind of did their about face, and they're like, oh, wait. Yeah. We need to use AI at every single turn. And at the time, I said consulting companies, for the most part, especially the mid tier ones, are gonna have to go out of business. The big ones are gonna be able to survive.
Jordan Wilson [00:33:38]:
They're gonna realize whether it's one or three years too late that they need that they need to become absolutely AI native, and, infuse AI in every part of their workflow and probably even change their pricing structure to the traditional, you know, build hourly rate to some sort of outcomes based pricing. But regardless of that, it is, I won't say shocking, but it's telling here that OpenAI and Anthropic are basically saying, nope. We're coming in with our own firms even though they have big partnerships, with the, you know, kind of the big fours, of the world. It's pretty telling here when OpenAI and Anthropic are coming in because one of the things that they're seeing is, well, companies need more help, implementing these, AI models. Because as capabilities race so quickly ahead, right, this is one of the biggest the the the the capabilities problem is one of the biggest problems that we've seen probably in generations. Right? Because, how companies are actually using AI models, and the models capabilities themselves are not aligned. Right? For the most part, even larger enterprises are maybe only using 10 to 20% of the AI model capabilities, and they just can't keep up or sometimes even understand everything that they have to offer. Alright.
Jordan Wilson [00:34:53]:
That's it for our big AI news stories, but let's quickly go over our bullet point round up at the end of what's new and what's next. Some of these are rumors. Some of these are smaller news stories, but they didn't make the big AI news story cut. But in any other week, they could be the biggest story. So pay attention. We're gonna go through them quick. Here we go. So OpenAI released three real time voice models, enabling natural conversation translation in instant transcription.
Jordan Wilson [00:35:22]:
This one y'all is big because, OpenAI did say that they're gonna be rolling these out, to ope, sorry, to chat g p t and maybe even to codex as well. Could even be as soon as this week. So yeah. Finally, the, you know, voice modes that we all use might actually be really good pretty soon. Google rolled out Gemini 3.1 flashlight. It's fastest and most cost efficient Gemini model yet. Anthropic released Claude across Excel, Word, PowerPoint, and Outlook in beta soon, so they rolled everything else in, rolled everything else out in general availability. So, next, Google DeepMind London staff voted to unionize opposing military and surveillance AI use.
Jordan Wilson [00:36:04]:
This happened after the goop Google DeepMind, kind of partnership with the federal government that we just talked about. Spotify rolled out support for saving and playing AI generated personal podcasts in users' libraries. So, yeah, if you've been wanting your open claw to create a daily podcast for you in Spotify, now you have that option. Leaked screenshots show the codecs app from OpenAI, control to be rolling out soon. So you may be able to control, codecs finally from whether it's the chat g p t app or a dedicated codecs app. We'll see soon, but it should be rolling out here any day. Week two of the Musk OpenAI trial revealed a lot, but, they had the, the Brockman diary explosive text in executive testimony. No real big new news this week, but I'm guessing that'll change this week as week three kicks off.
Jordan Wilson [00:36:57]:
XAI is preparing Grok Build, their new cross platform desktop coding assistant. Speaking of Grok, they released their, connectors in their mobile app. Google launched a unified health app and an AI health coach with new Fitbit, Fitbit Air tracker. Anthropic previewed dreaming for Claude managed agents at their dev conference, adding self improving memory and orchestration tools. Most of what they announced at their dev day, more on the back end inside of their console, so didn't think it was was too important for most of our audience, but we'll see when those things or if those things roll out to the front end, Claude. Adobe launched AI powered PDF spaces, turning Acrobat into an interactive content and collaboration workspace. OpenAI rolled out limited preview of GPT 5.5 cyber for critical infrastructure cybersecurity defenders. So kind of, not their version of Mythos per se, but kind of their version.
Jordan Wilson [00:37:58]:
Right, their next new powerful model, specifically, that's really good at cyber. Google AI Studio released edit mode, enabling direct UI component edits and image swapping via nano banana. I've been loving that. If you are a builder, make sure you check that one out. Meta is reportedly nearing internal testing of their new always on in, agentic AI assistant called Hatch, which is kind of being marketed externally as their open claw competitor. Google is testing Gemini agent mode, a new version of it, adding a dedicated tab for advanced automation features. Alright. And we have Google's IO conference here in, like, less than two weeks, so expect a lot of new Google announcements both next week and the week after.
Jordan Wilson [00:38:43]:
OpenAI launched codecs Chrome extension so good, enabling background browser automation across tabs. We did cover that in our new Friday features show, so just go back one episode where we dove into that. Microsoft released Copilot Cowork on mobile. We cover that in Friday's show as well. And Perplexity rolled out its new Mac app with built in personal computer support. Alright. That is a wrap. Technically, really busy week in AI news.
Jordan Wilson [00:39:12]:
A ton happening, a ton going on. So if you missed any of it, don't worry. That's what our Monday shows are for. But if you really wanna stay on the cutting edge, make sure you tune in well every day. Maybe this is your first time listening. So if you if that is you, thank you for listening. Make sure to subscribe to the show on the, you know, whether you're listening on Spotify or Apple Podcast, we appreciate this, but we do it Monday through Friday. So thank you for tuning in.