293 | The gloves are off in the AI implementation race and it will reshape the job market, Elon Musk had a “spy” inside OpenAI’s board, Claude for small businesses is now available, and more important Ai new for the week of May 15, 2026

Secure your spot for the MULTI-AGENT ORCHESTRATION AI COURSE: https://multiplai.ai/multi-agent-orchestration-course/

What if the biggest tech layoffs in history aren't really about AI… but the AI wave that justifies them is just getting started?

Tech giants are posting record revenues while cutting tens of thousands of jobs. CapEx is hitting historic highs. And behind closed doors, the courtroom drama between Elon Musk, Sam Altman, and Microsoft is revealing things that sound straight out of a Hollywood script.

In this weekend news episode, Isar Meitis breaks down the most important AI stories of the week — from the dissonance between booming tech revenues and mass layoffs, to the explosive testimonies in the Musk vs. OpenAI trial, to the cybersecurity exploits that just broke Apple's most advanced hardware in 5 days.

You'll get a clear-eyed look at what's actually driving the layoffs, who's winning the enterprise AI race, and why the next wave of disruption is closer than most leaders realize.

If you want to understand where the AI economy is heading — and what it means for your business, your job, and your industry — this episode connects the dots in ways most headlines won't.

In this session, you'll discover:

• Why tech companies are laying off thousands while breaking revenue records
• The truth behind "AI-driven" layoffs (and what Gartner's new data reveals)
• How OpenAI, Anthropic, and Google are racing to dominate enterprise AI deployment
• The shocking revelations from the Musk vs. OpenAI trial — including hidden personal ties and alleged lies at the top
• How AI just cracked Apple's most advanced security in only 5 days
• Microsoft's research showing AI agents corrupting documents on long tasks
• Google's new Gemini Intelligence layer for Android and Chrome — and what it means for the future of apps
• Anthropic's first major push into the small business market
• How AI-generated political content is reshaping elections (LA mayor race case study)
• Why xAI's Grok 9 and "Grok Build" could disrupt the Claude Code vs. Codex race
• The Figure AI robots that just sorted 28,000 packages in 24 hours with zero mistakes

About Leveraging AI

• The Ultimate AI Course for Business People: https://multiplai.ai/ai-course/
• YouTube Full Episodes: https://www.youtube.com/@Multiplai_AI/
• Connect with Isar Meitis: https://www.linkedin.com/in/isarmeitis/
• Join our Live Sessions, AI Hangouts and newsletter: https://services.multiplai.ai/events

If you've enjoyed or benefited from some of the insights of this episode, leave us a five-star review on your favorite podcast platform, and let us know what you learned, found helpful, or liked most about this show!

About Leveraging AI

• The Ultimate AI Course for Business People: https://multiplai.ai/ai-course/
• YouTube Full Episodes: https://www.youtube.com/@Multiplai_AI/
• Connect with Isar Meitis: https://www.linkedin.com/in/isarmeitis/ 
• Join our Live Sessions, AI Hangouts and newsletter: https://services.multiplai.ai/events

If you’ve enjoyed or benefited from some of the insights of this episode, leave us a five-star review on your favorite podcast platform, and let us know what you learned, found helpful, or liked most about this show!

Transcript

0:00

Hello and welcome to a weekend news episode of the Leveraging AI podcast, the podcast that shares practical, ethical ways to leverage AI to improve efficiency, grow your business and advance your career. This is Isar Matis, your host, and we have some very interesting things to talk about today. Today we are going to talk about some very big topics. The first one is going to be the big dissonance between the layoffs in the tech world compared to the huge success of companies in the tech world, and how that connects or does not connect to AI. And we'll hear different opinions as well as, more importantly, facts that support both sides of the story. We're gonna dive into that because I think that's gonna be a very big component of the future. I've been talking about this for a while, And I think it is really important we dive into this. The second topic is gonna be the soap opera of the trial between Elon Musk and Microsoft and OpenAI, Sam Altman, et cetera. A lot of things happened in the past week and a half that we need to cover, and there's really some interesting stories that came behind the scenes that I don't think were ever available before with a lot of interesting things that were said or written, and really crazy stories That can make a very successful miniseries in the future. We're going to talk about cybersecurity. We're long-running agents actually working properly, or are they putting your data at risk? We're going to talk about Anthropic's new direction towards small businesses and a lot of other interesting stuff. So let's get started. The first topic, as I mentioned, has to do with the big dissonance between how successful tech universe is right now versus how many layoffs are happening, and we're seeing even more of that this week. So companies like Cloudflare and Cisco and Meta all are laying off more employees. If you want the total numbers, since January of 2026, more than 150,000 tech jobs in the US were canceled. That is about 164 people per day that lost their jobs since the beginning of the year, a lot of them using AI as the reason/excuse. So let's dive into the actual numbers. Cloudflare revenue for Q1 was $639.8 million. That's 34% growth year over year. They ended with free cash flow at eighty-four point one million, simultaneously announcing, 1,100 layoffs. That's 20% of their 5,500-person workforce. Again, this is 20% of the people that are getting cut on a 30% increase year over year in revenue. By the way, this is the first mass layoff in the 16 years history of the company. This restructuring is going to cost them between $140 and $150 million per their own projections, and their stock fell eighteen to twenty four percent, depending if during hours or after hours on this announcement. So again, think about a thirty four percent growth and a eighteen to twenty four percent decline in the stock value. We're gonna talk about the potential reasons in a minute. Cisco has announced a fifteen point eight billion revenue. That's twelve percent year over year for a multi-billion dollar company, that is very significant. And they announced layoffs of 4,000 people. That is less than 5% of their global workforce, but it's still 4,000 people that are losing their jobs. Their stock went up sixteen percent despite the fact this quote-unquote restructuring is gonna cost them about a billion dollars in severance pays and so on. Meta also announced another restructuring in a very long list of restructurings. Their revenue crossed two hundred billion for the first time. Q4 net income was twenty-two point seven billion, and they announced another set of cuts of eight thousand employees. That is ten percent from their global workforce, plus canceling six thousand open positions that they had in the market for a total headcount of fourteen thousand positions. At the same time, they committed a hundred and fifteen to a hundred and thirty-five billion AI CapEx budget for twenty twenty-six, more or less doubling the budget of twenty twenty-five CapEx investment. If you want to add to that things that we reported before, Amazon cut sixteen thousand corporate roles while AWS posted a twenty-four percent growth year over year. Oracle eliminated thirty thousand positions even though they've seen record numbers, and so on and so forth. So the direction is clear. The tech market is booming. They're hitting record revenue and across more or less every parameter, they're breaking every record they ever had or at least doing very, very well with very significant growth year over year, and still they're laying off significant portions of their workforce Now, if you want to hear what the actual CEOs of these companies said, Chuck Robbins, the chair and CEO of Cisco said, and I'm quoting, "The companies that will win the AI era will be those with focus, urgency, and discipline to continuously shift investments towards the areas where demand and long-term value creations are strongest." Again, that says nothing about people or employees, but it says that the investments needs to shift and shift fast. Matthew Prince, the co-founder and CEO of Cloudflare said, "Today's actions are not a cost-cutting exercise or an assessment of individuals' performance. They are about Cloudflare's defining how a world-class high-growth company operates and creates value in the agentic AI era." Again, if you think about the underlying of what it says the same thing. We are rethinking how our company works and structured, and we know we will need to continuously do this to be aggressively growing in the new AI era. Now, diving a layer deeper, one of the things that Cloudflare internal transformation shared is that their internal AI usage grew over six hundred percent in three months. This last quarter, they grew their AI usage six folds. Or as their CEO explained it, it's like going from a manual to an electric screwdriver. This is very significant when it comes from a CEO of a company that is trying to grow in this era that is saying it's going like from manual screwdriver to an electric screwdriver. And it explains, I think in a very good way because most people have done both, how significant is the AI change. Now, in Meta's example, they are not just trimming headcount, they are completely restructuring how their org chart looks like and how it operates. Entire teams are being reorganized into what they call pods under the new applied AI organization. So there are different new role categories such as AI builder and AI pod lead, which are replacing traditional function-based titles. This means, again, it is not a let's make the company more efficient because now we can do this better. It is, again, let's completely rethink how a company works in the AI era and try to build something new that will serve the new needs of this era. Another really interesting parameter in this whole conversation came from Bloomberg this week, And they suggest that half of AI-attributed layoffs will result in roles being rehired offshore or at lower salaries. Meaning while companies are laying off a lot of US employees claiming that it's due to efficiencies, some of those efficiencies, potentially half of those efficiencies, will come not from AI, but will come from hiring people at lower salaries, either in the US or offshore. Which kind of again shows that nothing has really changed in the corporate world other than the fact that they now have another excuse to use, or maybe that is not the case. So let's keep diving deeper, and then I will try to answer what I think is actually going on. So another thing that we talked about last week and in previous weeks is that hyperscalers' CapEx is now between six hundred and fifty to seven hundred billion dollars just this year alone. So Google, Microsoft, Amazon, Meta are collectively committing to over half a trillion dollars in this single year alone to grow their compute and their capabilities from a CapEx perspective. At the same time, we are seeing token consumption inside enterprises skyrocketing, and that's across the world. There are several different companies who are tracking this information. The number one use case is still writing code, but this is diversifying, and I'm sure that we're going to see in the second half of twenty twenty-six a growing consumption across other use cases that will slowly take a bigger and bigger share of the pie while the overall pie is keep on growing at a crazy pace. So what do we know? So far, we know that these companies are laying off a huge amount of people. We know that their revenues are skyrocketing, and we know that they're investing crazy amounts of money in compute or in infrastructure that is supposed to trickle down and be available for them to sell more of their AI services to other companies in the future. So what is the limiting factor? What is going to make this not successful? What is going to make this AI transformation slow down dramatically? Well, the lack of knowledge. I've been screaming that for a very long time, and this is why I've been focusing on delivering AI workshops and AI courses to a huge and wide range of organizations across the world, from small businesses all the way to Fortune 500 companies. And the reason for that is I think teaching people how to use AI, teaching leadership how to structure their organizations around AI is going to be as important, and in the long run, more important than the progress in the technology. And the reason I'm saying that is because I think we reached a level of maturity of these models that are now good enough to do, well, more or less everything. I'm telling you right now that we're going to talk later on in this episode about a research by Microsoft that says otherwise, at least in the current models. So we're gonna get to that in a minute or in a few minutes. But it is very obvious that the biggest gap right now is how to implement AI successfully and not are the models good enough. And this connects to something we started talking about last week and the week before that, and is even growing this week, which is the deployment of the main leading companies in the AI space, so OpenAI and Anthropic and their partnerships with the leading consulting companies in order to deliver AI capabilities straight into the consulting companies, as well as their recent partnerships with private equity companies in order to deploy advanced AI capabilities into the private equity's portfolio companies. Well, OpenAI made another announcement this week of starting their own company that calls OpenAI Deployment Company or ODC for short. That's not a very unique name, but for a company who named their product ChatGPT, I'm not surprised. This new company that they just launched has a $10 billion pre-money valuation with over $4 billion in investment. OpenAI themselves contributed $500 million in equity. So if you want, this is a separate company from OpenAI that is going to focus on implementing of OpenAI's technology inside of companies Now, to tell you how aggressive this move is, OpenAI has more or less promised or guaranteed a seventeen point five percent annual return to the companies who invested in this company. That is really, really high for a company that doesn't exist yet and just exists on paper. Now, on the very first day of its existence, they acquired A UK-based applied AI firm called Tomorrow, spelled T-O-M-O-R-O. And this company has one hundred and fifty forward deployed engineers that work with different companies already, and they have a track record of deploying new production capabilities within twelve weeks or less, and have done it in companies such as Tesco and Virgin Atlantic and Supercell, so really large companies. As part of this move, we learned another interesting thing, that OpenAI's enterprise revenue now exceeds forty percent of the OpenAI total revenue. And if you remember the code red and switching from trying to do everything for everyone to let's focus on what matters, which is enterprise revenue. So it seems to be definitely working for them. And Brad Lightcap, the CEO of OpenAI, summarized it very beautifully. He said: "Our customers tell us they need help going from pilot to production. Deployment company will put our engineers inside their teams with resources to ship." If you remember, we talked about the fact that Anthropic launched something very similar roughly at the same day. And now we've learned this week that Google is also jumping all in into this race, that they have established an ecosystem of three hundred and thirty thousand consultants working across the big consulting companies They're going to co-invest together with those consulting giants, so companies like McKinsey and others will be training all their people on how to effectively deploy Gemini models into enterprise companies in order to gain as many benefits from AI as possible. Thomas Kurian, the CEO of Google Cloud, summarized it well. He said, "We're at a pivotal moment where AI fundamentally defining what's possible for every business. By bringing together McKinsey's strategic guidance with Google Cloud's AI platform, we're helping organizations reimagine entire value chains and build the agentic enterprise that will define the next era of global industry." So slightly different approaches from OpenAI versus Anthropic versus Google. They're all aiming at the same thing, accelerating adoption of AI in an effective way in the largest organizations in the world. But now I want to give you the final piece of information that will make the question mark even bigger. So Gartner just shared a study that finds no correlation between AI-driven layoffs and higher ROI. So a Gartner study of three hundred and fifty global business executives at companies with at least one billion of revenue found that while eighty percent of those who piloted AI or autonomous technologies reported workforce reductions, there was no statistical correlation between those cuts and higher returns on investment. The flip side, on the other hand, was true. The highest ROI companies amplified workers rather than replacing them. So let's do a quick summary, and then I'll tell you what I think is happening and where I think this is going. And again, I don't have a crystal ball, but I'm looking at this every single day and thinking about this a lot. The first thing, the productivity paradox is real, and it is accelerating, right? So we have companies who are growing very, very fast and on the other hand are laying off a lot of people. We have a bottleneck in deployment right now. So the biggest problem in deploying AI right now is not the technology, but it is the knowledge on how to do this effectively. As I mentioned, I've been doing this with companies for the last three and a half years, and I see this very, very clearly. Most companies, regardless of size, regardless of industry, just do not know how to implement AI effectively and we are seeing the biggest investment maybe in history to solve that problem. The Gartner data point is very interesting, and again, not surprising to me that they're showing that laying off people is not the right way to go, but rather to amplify people, teach them how to use AI effectively, that in order to gain as many benefits as possible. Speaking about that, I am actively doing workshops to companies on how to build effective and safe multi-agent orchestration solutions across their organizations. I've already done two of these workshops, and they were extremely successful, and the companies who have done this within a couple of weeks are already deploying production-level solutions that are making dramatic changes in their processes. And so if you are looking to do something in your company, please reach out to me either via email or LinkedIn or any other method. And if you're an employee somewhere or a solopreneur, I teach courses, and we currently sold out the first three cohorts of this course that teaches the same thing as an online course, but we're now selling seats for the August course, And you can find a link to the course in the show notes. Again, these are selling really, really fast. We started selling them a few weeks ago, and we already sold the first three cohorts. So the next one will sell as quickly, and if you want to be a part of it, you should sign up right now. By the way, between now and the middle of June, there's $100 discount. It was $500 in the beginning, then $200. Now it's $100, and once that expires, We will not provide any additional discounts and probably will raise the price over time because similar courses are selling between thirty-five hundred dollars to five thousand dollars, and so the twenty-five hundred dollars that is the cost of this course is still a steal. But back to the summary and what I think and where I think we're going. Yes, there is different angles and different data that is showing that these layoffs are not a hundred percent related to AI. It's either AI washing based on over-hiring previously, or the fact that there were, uh, zero interest and it was very easy to raise money in order to hire more people, or that the ROI is not really increasing when you lay off people, or that half of these cuts will be replaced by people in cheaper labor regions. All of that is true. I don't deny any of this. These are most likely facts. However, Cloudflare's internal AI usage surging six hundred percent in a single quarter is a fact as well. Cisco's two point one billion in AI infrastructure orders is real as well. OpenAI's enterprise revenue growing and crossing forty percent of overall revenue is a fact as well. The huge investments right now into helping companies implement AI is also very real. So what does this mean? This means that right now, in this very moment in time, maybe a lot of the layoffs that we're seeing are not directly correlated to AI capabilities. However, all the required ingredients for a massive layoff growing and continuing across a much wider part of the tech world and then the rest of the economy are there. The ingredients are existing. The tech capabilities are there, the models are there, the agentic capabilities are there, and the only thing that's missing is knowledge and learning how to do this effectively, and that is coming, and that wave is coming very, very fast. And like all the other stuff that we've seen so far, it is coming faster and at a higher scale for larger enterprises because that's where money is. But it will, over time, trickle into smaller and smaller businesses and for the rest of the economy. And while in the immediate future, and you heard me say this multiple times before if you've been listening to this podcast. In the immediate future, you can gain market share and grow by keeping your people and maybe hiring additional people to help with the AI adoption and restructuring correctly in order to benefit from AI and capture market share from your competitors who are not doing it at the same pace. But X number of months or years from now, that will level off. Using AI effectively will be like using Microsoft tools effectively or like using computers effectively, and everybody will do it. And there's a finite demand. That is a fact. Yes, demand is growing in specific things because it's becoming cheaper and more people will be able to afford different things. I agree with that as well. But I don't think it is at the same pace of AI being able to replace humans in doing specific jobs. And I think we are going to see an era where we're going to see massive layoffs that may level off after X number of time. But I think in the medium term, I don't see a way around this, and I really hope I'm wrong. The second topic that I want to cover today is the trial between Elon Musk and OpenAI/Microsoft, and we learned a lot of interesting things in the past few weeks from documents and testimonies that happened during that time. So we heard from a long range of people who testified this past few weeks in the trial. That includes Sam Altman and Ilya Sutskever and Mira Murati and obviously Elon Musk. But we also heard from other interesting people such as Siobhan Zillis, who has a very interesting background story that we did not know about that has a whole interesting layer that adds an entire layer to this whole story, and many others. So let's try to break this down and talk about some of the things we've learned. And again, I can't cover all of this because this will be like a whole series of episodes, not just one episode. I suggest you go and read the different articles about this to really learn what's going on. So I'll start with a few things that Sam Altman said that stood up to me. One, he said, and I'm quoting, "I'm sure there are some times in my life when I did not always tell the truth." Now, he said that without being prompted. There was no clear question about his honesty or ability to speak truthfully about his actions. He just went ahead and said that. Another thing that he said on the flip side, as far as Elon Musk's push that they have stolen charity, he says, and I'm quoting, "It feels difficult to even wrap my head around that framing. We created the largest or one of the largest charities in the world." And he obviously refers to the fact that the nonprofit part of OpenAI is now owning a big part of OpenAI, which now makes it most likely the best-funded charity ever in history Now, related to this topic, Ilya Sutskever, who was OpenAI's chief scientist at the time, also testified that Sam Altman "consistent pattern of lying," again, that's a quote, is something that is a big part of his character. Now, we all know that Sutskever played a very big role in the ousting of Sam Altman when it happened, and then there was a big silence about whether he's gonna stay in the company. We haven't heard about anything he was doing for a very, very long time, and then he left and started Superintelligence, a company that now he runs and that raised a crazy amount of money to pursue superintelligence. But he clearly testified that Sam Altman lied multiple times while being the CEO of the company. Mira Murati, the former OpenAI CTO, who did a video deposition, testified that Altman was, and I'm quoting again, "Creating chaos inside the company by saying one thing to one person and the opposite to another." She also said specifically that Altman lied to her about safety clearances for a new model and falsely claimed OpenAI Legal had determined that the model did not require a review by the deployment safety board. Now, on the stand, she was asked if she perceives Altman to be not candid, truthful, or honest. She said, "Not always." She also said that Altman undermined her as CTO and pitted other executives against one another. We heard that from other people in the past. On the flip side, the defense brought in testimonies that said that Murati was, and I'm quoting, "She was waiting to see which way the wind will blow." Basically talking about the fact that she wasn't willing to stick up her neck and choose a side in the whole process of the Sam Altman ousting. We know that she was a part of the process, but then she was waiting to see how things are going to go. Now, we also learned that while Mira Murati was a part of the people who were trying to get Sam out, he potentially did not know that, and he was texting with her back and forth as the whole thing was evolving and happening. So there's a whole text exchange between Sam and Mira Murati during that time where he's asking her what's going on, and she's giving him vague answers, but clearly telling him that he's on the way out and that the board is very clear with their direction to fire him. So Altman texted, "I'm still willing to just walk away if that helps." Murati replied, "They're convinced about their decision." And then Altman asks, "For me to be fired or some new thing?" Murati says, "Yes, for you to be gone." So she is one of the people who potentially initiated his ousting, and Sam trusts her to be his ears inside and during this process, and she's giving him the information that he's most likely on the way out. We know how that ended. He came back, he took the helm again, and then restructured the board. And then the two people who are most likely closest to the top, which is Mira Murati and Ilya Sutskever, both eventually left and started their own companies. But two really crazy, surprising things that at least I learned that I don't think I've ever heard before. Uh, one is that Shivon Zilis, who sat on the board of OpenAI until way after Sam left and came back, actually had four kids with Musk. So she's the mother of four of the many kids that Musks have, and she was sitting at the board of OpenAI reporting back to Musk probably anything he wanted to know about what's going on in the board, and she did not share that with anybody because she had an NDA with Musk not to share that they had kids together think about the plotline of every good soap opera, and you'd never heard anything like this, that one person has an NDA with another person on the fact they had kids together, and then in one of the biggest fights in business history, the one person sits on the board of the company that the other person is going after. It's, it's absolute madness. Now, the flip side, by the way, also became obvious in part of Zelis's, cross-examination, and that's the fact that Elon Musk approached Karpathy to bring him into Tesla when Elon testified exactly the opposite. He said that Karpathy approached him, but Zelis, after shown texts between her and Musk, admitted that was the case. Another thing that we learned that was interesting is that two days before the trial, Musk texted Brockman to gauge a potential settlement interest. When Brockman replied that they don't want to settle, but they can have both sides drop their lawsuits, Musk wrote back, "By the end of next week, you and Sam will be the most hated men in America." Now, the judge ruled this text inadmissible because AI submitted it too late for the trial. But again, this became obvious and became one of the things that the jury have heard that actually happened. We haven't heard anything in the trial that will make Brockman and Sam the most hated people in America. I don't know if that was a threat by Elon just to get them to maybe agree to whatever settlement he was offering or suggesting, or that he has some ace up his sleeve that he's gonna pull either the final stages or after the trial. But this, again, is something that came out to the open Now we heard a lot of other interesting things during this process, and I'm not going to dive into all of them, but here are a few interesting things. Altman himself as a person, not any one of his companies, hold a personal stake of over two billion dollars in third-party companies that are doing business with OpenAI, which obviously has a very serious conflict of interest. Per him, he excluded himself from any conversations that had to do with the relationship with these companies. But the fact that there are a few of them and the fact that the number is in the billions makes it a problem in my personal eyes. Again, I'm not saying from a legal perspective that makes a difference or not. We also learned that Microsoft almost did not invest in OpenAI in the beginning. So in August of twenty seventeen, days after OpenAI Dota two bot beat professionals, uh, in a specific task, Altman proposed a three hundred million Azure compute partnership to Satya Nadella. Azure chief Jason Zander pushed back and he said, "For those numbers to make sense, we'd have to be generating significant incremental revenue directly due to the deal that is five hundred million dollar-plus that could be gained in a more efficient way." So basically, they almost walked away from the deal with OpenAI. The other thing that we've learned is that to date, Microsoft invested between across cash as well as compute that they provided OpenAI about a hundred billion dollars to get their current share in the company, which as of right now shows up as a very successful bet because their shares in OpenAI are worth a lot more than that. And if they do actually survive the trial and they do actually go public, they're gonna be worth even more. But the other thing that we learned in this trial, which puts Microsoft in a very delicate situation right now, is that approximately forty-five percent of Microsoft remaining commercial performance obligations come from OpenAI. Basically, if OpenAI goes down, Microsoft future projections, almost half of them will disappear. That is not a healthy situation to be in for any company, especially not a company the size of Microsoft. So where does that put us other than the soap opera and what's going on? As I mentioned previously when the trial started, and again, I didn't think it would actually go to trial, but when the trial did start, I shared that on the podcast that if for whatever reason this stops OpenAI from going public or even delays it by a few months, the shit show that it is going to generate is going to be very, very significant because so many things are interconnected here, and so much money is involved, and so much future revenue of a very long list of companies is dependent on OpenAI going public and having the money to do the things that they said they're going to do that I don't know what will happen if that doesn't happen, but it is not going to be something fun to watch for a very, very, very long list of companies and investors and so on. But that can happen. Now, the way it will roll out is it is a jury trial, but the jury does not decide. They just recommend. The judge will make the final decision, but obviously the jury's suggestion or recommendation has a serious weight, and because it is a jury trial, this can go in any direction. And so I'm very curious to see where it's going to go. What is very obvious to me, and I'm not picking sides, I'm literally just stating what I've heard in the past two weeks. We have a person who is currently the CEO of one of the largest companies in the world, one of the most successful companies in the history of the world, and soon to be a most likely trillion-dollar-plus company, publicly traded company, that has two of his closest members of leadership in the company saying that he lied on multiple occasions and him admitting that that's the case on his own without being triggered to do so. I see that as seriously problematic, and I assume that both the jury and the judge will pay attention to that and potentially will allow everything to move forward, maybe not with Sam Altman as the head of that organization. I, again, I don't have a crystal ball. I'm not a legal person. I don't know how the proceedings are gonna turn out. But this is my gut feeling because it doesn't make sense to me, regardless of what Elon wants. It doesn't make sense to me that a person that is going to run one of the most influential companies ever to be a person that regularly lies to everybody around him in order to achieve what he wants, and I don't think a person like that should be the CEO of a publicly traded company. I will obviously keep on updating you as this goes forward. As I mentioned, there's a lot more you can learn from what happened in the case. So go and read some of the things that happened. There's more stuff to learn. I think I covered all the important stuff. Now we're gonna switch to rapid fire. There's a lot to cover still in that. The first thing is around cybersecurity. So openAI announced its EU Cyber Action Plan on May 11th, offering vetted European cybersecurity teams and governments and the EU institution, for AI to get access to their GPT 5.5 Cyber, which is their model that is purposely built to do red teaming and find loopholes and risks in different software platforms around the world. If you remember, OpenAI announced it just shortly after Anthropic shared that Mythos, their new version of model, is highly problematic when it comes to cybersecurity. now, by the way, while Anthropic very first move when they announced it was to create what they call Project Glasswing, which has forty plus companies in the US that got access to this, they were approached multiple times by the EU Commission to allow them access to the model as well, and they declined even after OpenAI gave them access. So we have two companies with two different approaches on how to allow or not allow access to advanced models that elevate the cybersecurity risk dramatically. But still on the same topic, A cybersecurity firm called Caliph, which is a company from Palo Alto, used Anthropic Mythos Preview, which is the model they shared with this company as part of the companies they shared it with, were able to execute the first ever macOS kernel memory corruption exploit using Apple Memory Integrity Enforcement. So it's a component inside of the M5 silicon hardware that is built and deployed and is the most advanced hardware that Apple generates. And using Mythos, they were able to get access to critical components of the actual chip of that runs all the most advanced Macs in the world right now. Now, to be fair, Mythos did not do this on its own. It was a collaboration between the employees of Caliph together with Mythos. But what they said is that the overall exploit from the very research stages all the way to having a working exploit took about five days. They're saying that without Mythos, they might have been able to do this, but it would have take them multiple months. So that gives you an idea on how powerful this model in the right hands to get access to things that are the most sacred gardens, if you want, in the universe, because Apple has always put themselves as the safer solution that builds their own infrastructure that nobody can touch and get access to. And now within five days, a company was able to get access to their most advanced chips. The people from Caliph made two interesting statements about this whole incident. One is Apple built MIE in a world before Mythos Preview. Basically, they're saying the rules have changed. They've built the best, most secure system possible, but that was before these kind of models existed. That's very, very scary because, again, if they were able to break into this, think about everything else that was built pre-Mythos Preview being available to them, which is basically every piece of software and hardware we know in the world. The other statement that they said is, "We're about to learn how best mitigation technology on Earth holds up during the first AI Bugmaggedon." And I assume they mean that they see a huge wave of issues and software collapsing and cybersecurity problems as a result of these kind of models being available in the world. Now, how is that going to evolve? How many more companies are gonna get access before they release this to the public, and so on? I don't think anybody knows, but it is obvious that we are at a turning point, not just from an industry learning how to use AI, but also from the level of risk that AI generates to cybersecurity. The next topic that I wanna talk to you about is a very interesting study that was done by Microsoft. They just found that AI agents running on very long tasks are actually completely corrupting and/or deleting the documents that they are working on. So what they tested was three frontier models, Gemini 3.1, Claude Opus 4.6, and GPT 5.4, so one version before the last of all the most advanced models. And what they found, and I'm quoting, our findings show that current LLMs introduce substantial errors when editing work documents with frontier models Gemini 3.1 Pro, Claude 4.6 Opus, and Gemini 5.4, losing on average twenty-five percent of documents content over twenty delegated interactions and an average degradation across all models of fifty percent." What did exactly Microsoft researchers do? What they did is they created a benchmark simulating a multi-step workflow across fifty-two professional domains. So some of them are very obvious, but some of them including and music notation, right? So things that are not from the business world, and that in many cases are far more complex than just solving spreadsheet tasks, which is what most people are using it for. And then the test requires models to split, reorganize, and merge documents across twenty different interactions. So not your day-to-day tasks, but bigger, more complex, long-lasting processes. And what they found is that equipping LLMs with tools, so basically going more agentic, actually made the situation worse rather than making it better when it comes to the final outcome. They did find that better, more advanced models are slightly safer in doing this and delete and destroy less of the content. But maybe the most important findings is that the stronger models delayed the failures to later rounds in the process. What this means is that most of the testings that most of us and anybody who's testing this are doing today are not a good prediction to how that's going to extrapolate once you generate more complex multi-step tasks. Basically, if you tested it on five steps and it worked perfectly, and you instead test, tested it on nine steps and it worked perfectly, it still doesn't mean it's gonna work perfectly over 20 steps because these good models were able to delay the failures to later states in the process. What does that mean to all of us? I think very little in the short term. I, again, do this for a living, I do this for myself, I do this for many different companies, and the vast majority of business use cases do not require a 20 steps reorganization of entire files and sets of data. They require much shorter steps. But what you can learn from this, which I strongly support and I build into every single process that I create for myself and for other companies, is human test points and checkpoints in critical steps of the process. If you do that, then you control what's gonna be the outcome, and you don't have to verify it in the very end. You create checkpoints in the middle, you see what's happening. If you don't like it, you comment, and then you go back or you continue with your comments to the next step, allowing you to navigate the overall AI process versus allowing it to run completely independently on its own. Now switching from Microsoft and this research to a big announcement from Google that is both important and surprising at the same time. So Google introduced what they call Gemini Intelligence, which is a new initiative that basically transforms everything you know from Google as far as operations. So Android and all of its components, as well as Chrome and all of its components into a unified intelligence system that Gemini controls in the back end So this is not a new set of features or upgrades. It is being positioned as an intelligence layer underneath Android itself. So Google is saying that Gemini Intelligence will be able to move across apps, understand what is on the screen, and complete tasks that would normally require the user to jump across multiple services. So what are some of the features? App automation, which is what I just mentioned. it allows Gemini to understand what you're trying to do and go across multiple tools that you have on your phone and complete the task almost all the way to the end. So the examples that they gave is Google can find a class on the syllabus on Gmail, then identify required books from other sources, then placing them in your shopping cart on another platform, all without you having to do anything. And all it's going to wait is for you to do the checkout step, which is not allowed to do on its own. They're also introducing Chrome Auto Browse, which is supposed to be available at late June, so a month from now-ish. And Gemini in Chrome will be able to summarize articles, compare content across multiple sites, auto-browse on its own, and complete tasks across multiple tabs that are available in the browser all independently. They're also introducing Rambler, which is a new voice-to-text engine that is aware of the context and your personal preference and will be able to correct every double words or uhs and ums that you're saying and stuff like that, and be able to come out with crisp, well-defined text every time while you speak. I find this really interesting because I voice type more or less everything I type, so having better voice typing capabilities would be fantastic. And I'm an Android user and I use Chrome, so this sounds, uh, perfect to me specifically. It also supports switching languages in the middle of the sentence. Me as a multilingual person, that is great as well. They're also introducing something that they call Create My Widget, where you'll be able to explain in simple English what you want the widget to do, and you will be able to get the information, create the user interface, and give you the widget. This is initially coming for Android, but then for all the wearable OS as well, so you'll be able to create new widgets for your Android watch as well. I was talking about these kind of things about three years ago when I said I think the days of the app stores are not very, very long because I think the concept of app store will cease to exist because anybody will be able to create whichever apps they want on the fly as they need them. And I think this is just the first step in that direction. Another interesting feature is Smart Auto Fill. So if you think about it today, you go to Google and you open Chrome, and there's a form shows up, and it knows how to fill up your name and your address and stuff like that. Well, based on your personal information and based on the Gemini intelligence, it will be able to fill up more or less the entire form for you in a much more intelligent way. Now, they also introduced Google Book, which is their first new wave of laptops instead of Chromebooks, that is supposed to have this Gemini intelligence as the underlying layer that runs everything. And they described it as, and I'm quoting, "The first laptop designs from the ground up for Gemini intelligence." Now, they shared that they invested a lot in the privacy aspect of this, which is not surprising. But to me, the most interesting thing is the timing of all of this. So if you remember, we talked about the fact that Apple Intelligence is going to release its first, maybe first working version that is gonna be powered by Gemini. This will happen in the WWDC conference that they're doing every single year that is just around the corner. So they're now announcing a competing technology to that, that they're releasing on their own, and they're announcing it ahead of their partnership with Apple. But to make this even more interesting, next week is Google I/O. So on the nineteenth of May, Google's gonna take one of the biggest stages in the world and share all the new news about everything Google is doing. Now, they could have waited a week with this news, right? It's just less than a week away, and they could have announced this as part of Google I/O. That tells me that something bigger is going to be announced in Google I/O. Otherwise, this would have been the biggest news, right? Because you need to ask yourself then, what is bigger than reimagining your entire system, the world's most popular mobile operating system and the world's most popular browser. You're announcing this ahead of your biggest event. What are they going to announce in actual Google I/O? I am very, very curious, and I will obviously keep you updated next week. I told you in the beginning we're gonna touch on Anthropic that is finally catering to small businesses as well. So we know very clearly from the crazy meteoric growth of Anthropic that they've been focusing very much on enterprise, and we know that that has led to the code red in OpenAI to do the same thing. Well, Anthropic just announced something really exciting. They announced what they call Claude for Small Businesses, and that's a package of connectors and ready-to-run workflows designed specifically for small businesses, and it's going to embed many tools that many small businesses use and connect it directly to Claude and the Claude working environment. I'm personally very, very excited about this because, again, I'm a small business, but also because a lot of the clients that I work with and a lot of the people that I teach in my courses. So what are the tools that are going to be connected? They include QuickBooks, PayPal, HubSpot, Canva, DocuSign, Google Workspace, and Microsoft 365. And they announced it together with a long list of workflows that are now available by just simply asking for them in English or by forward slash commands inside Claude Code and Claude Cowork. And they're enabling things such as handling payroll planning, monthly analysis, closing the month from an inventory perspective, invoice chasing, uh, campaign management, and many other aspects that small businesses do regularly and usually are not having enough people or enough capacity to handle all of them well. Again, I haven't tested it. I don't know how well it works, but based on everything else I've seen from Anthropic so far, this could be life-changing for small businesses. Now, why is this critical on a larger scale? Well, small businesses today account for forty-four of US GDP and employs nearly half of the private sector workforce. And as I mentioned, so far the focus of all the AI labs have been large enterprises or single individuals, and it's the first time that there's a very clear focus on small businesses, which I find exciting. And now that they've done this, I'm pretty sure that OpenAI and potentially other labs will follow as well. Another thing that happened this week that I have been talking about for a very long time, but is now taking a completely new shape, is that Spencer Pratt, who is running as the mayor of Los Angeles, he's a former reality TV star, has created a video for his campaign that is all AI-generated. It was created by filmmaker Charlie Keiran, and it portrays him as a cinematic hero-style Batman that is fighting Mayor Karen Bass and her staff and other known left-wing characters from California and other in the US, and it also includes, in addition to these characters, other known figures from both sides of the aisle, portraying a very clear line between the left and the right and showing him as Batman fighting for justice against people who are cruel and don't really care about the suffering of the people. If you haven't watched the video, go watch it. First of all, it's really entertaining, but more importantly, it will show you what is possible today in the AI space and how that can be impactful in whatever political races we're going to see in the future. I said that multiple times. I think the combination of the ability to generate any kind of reality very easily with AI is a huge problem across the board, and one of the immediate concerns is obviously the elections. And now we're seeing this on the LA mayor elections, but we're very close to the midterm elections, and I am certain we're going to see a lot of AI being used over there as well. It just provides the ability to deliver messages in a way that was not possible before. This particular v-video got millions and millions of views very, very quickly that I am certain that Spencer couldn't have gotten any other way. Now, are all of these people who watched the video related to the LA mayor elections? Of course not. I'm one of the people who watched it, and I live in Florida, so obviously I'm not gonna impact that race. But it shows you the power of AI-generated videos and/or images. We've seen Trump use multiple memes that are AI-generated to deliver messages, and I'm sure we're going to see more and more of this. One of my earliest episodes of this podcast, and you can scroll back to the first ten or the first 20 episodes, you will find an episode called "The Truth Is Dead," and you will see that I anticipated that this is happening, and I'm terrified of what might be the implications of this on society as a whole and definitely related to elections A few interesting news from OpenAI. First of all, OpenAI announced that GPT mobile app will now have Codex built into it, so it will provide users access to what is currently running on their Codex platform on their computers, but be able to control it and interact with it and engage with it from their mobile apps, very similar to how remote works for Claude. So again, not a new feature, just a new feature in the OpenAI universe to close a very big gap that they had with Claude. Now, to be fair, I don't know how many developers will actually use this from a development perspective. You usually wanna do this in front of your computer. But as a very heavy user of this for other use cases myself, I interact with Claude Code and Claude Cowork through my phone almost every single day, and either create or continue working on entire new workflows and automations and teams and agents, uh, from the phone. And so if you are in the OpenAI universe and you travel a lot, or you just do your day-to-day routine, and you take your kids to their classes and walk the dog and, uh, do stuff like that, it will allow you to continuously run and improve and progress the things you're working on straight from your phone. We talked about OpenAI push for implementation with their new partnerships and their new owned deployment company. Well, they are also now promoting very aggressively companies to start using Codex instead of other platform, and on a post they put on X, they are actively telling people to go and tell their CTOs to take a look at this particular post. And what the post is saying, and I'm quoting: "Want to officially use Codex at work? Send this post to your CTO to bring your team to Codex. Eligible enterprise customers who switch in the next thirty days gets two free months of Codex usage for new users." Now, if you think about what does that do, it takes companies who are not completely certain about where they wanna go and have been testing different applications to be able to get two complete months across every new user. That is a huge amount of money that could be saved to an organization, and that's a very aggressive move by OpenAI to capture more users away from Claude and other tools. And speaking of the competition in that space, two very interesting things happened this week. Microsoft started canceling most of its internal Claude Code licenses and transferring and transitioning thousands of developers inside of Microsoft to GitHub CLI. Now, while the company is framing this as a strategic move to focus on its own products, which makes perfect sense, it's pretty embarrassing that you're trying to sell a product and your internal employees are using the competitor's, uh, product. And this was actually met by a big frustration by multiple developers inside of Microsoft who said That they should be using Claude Code because it delivers better results. But the other interesting piece of news is that a lot of people are saying that this is not really a strategic move, but actually a way to save a lot of money ahead of their end-of-year final report. Microsoft ends their fiscal year at the end of June, And if they're going to cancel a huge amount of Claude licenses and tokens, they can show that as savings on their P&L at the end-of-the-year report to potentially drive the stock a little higher. raising the chip stack in this bet for enterprise developers is xAI, now a part of SpaceX They just finished training their Grok version nine model that has one point five trillion parameters on their new set of Blackwell GPUs. They anticipate this model to be a gigantic leap forward over the existing capabilities of the existing Grok model. And in parallel, they have, by mistake or not totally by mistake, have deployed a leaked user interface on the Grok web version that shows what they call Grok Computer button. Which led to a whole set of rumors. Some of them are probably true, some of them I'm not a hundred percent sure. But It seems like they're going to launch what's gonna be called Grok Build, which is a desktop coding application that will compete with Claude Code and Codex. And based on the latest information, it has all the different capabilities, literally feature parity with those two leading labs' existing models, including support for plug-ins and MCPs and skills and connectors, and enable Git tree integrations and local file management and web browsing and developer server spawning. Basically, all the things that Claude Code and Claude Cowork can do, it will be able to do. Will that put Grok and xAI back on the map and on the race that they kind of fallen out of? I don't know yet. It will be very, very hard for them to do this because I think there is a massive level of adoption right now for Claude Code and a growing adoption for OpenAI, and it will be hard to get market share out of that. But with the right pricing and/or the right approach, they might be able to at least bother the other two players and get some market share. And I will report as we learn more on this when it's finally deployed. And after all of our conversation in the beginning of this episode, as far as efficiencies that are coming and how it changes everything, I want to share something from the robotic world. Well, Figure AI, which is one of the leading companies in the world that are developing humanoid robots, just did a very interesting experiment. Their humanoid robots, three of them actually, that are named Bob, Frank, and Gary, were sitting and going through packages, trying to sort through them, through different types of packages, and they were trying to get them to run for eight hours completely independently without stopping and without human assistance and without making any errors. The reality was that they did this in a live stream, and once they finished the eight hours, they decided to just keep on going, and they were able to go for twenty-four hours. So these three robots were able to sort through twenty-eight thousand packages in twenty-four hours, hu-- operating at human-level speed while making zero mistakes. Now, is this completely realistic? I'm not a hundred percent sure. I don't know exactly what was the inputs and the outputs and how the test was set up. But the reality is we are now having robots who can do pretty sophisticated tasks in the real world that humans perform so far at human level, which again, these robots will improve every few months, guaranteed, and humans won't. But right now it's at human level, but without making any mistakes while running practically nonstop. So while we are seeing the impacts on the software knowledge work first, the robotic wave is following. It's not very far behind. Again, this, there's a very big difference between this test and doing it at scale, and there's a lot of other questions from, uh, security and safety of other employees, and there's a lot of other questions. But these will get resolved in the next two to three to four to five years, and then robots will be able to do any human labor that humans are doing from a blue collar job, or maybe not every, but a big enough portion to completely disrupt that part of the economy as well. So on that note, which again, I'm not sure if it's positive or negative, we might come out of this with a lot more free time and being able to do things that are a lot more human and having a lot more freedoms and getting access to a lot more services and goods at a much cheaper rates. I don't know how this plays out, but I think this is where we are, and I think this is where the trajectory is going, and all I'm trying to do is to keep you informed. That's it for today. I hope you have an amazing rest of your weekend. If you wanna learn more about the courses and the workshops that we deliver, just go to Multiply spelled with AI in the end, so M-U-L-T-I-P-L-A-I.ai, or just click on the links in the show notes. You can do it right now. And if you are in the show notes and you're looking at your phone right now or your computer or however it is you're listening to a podcast, I would really appreciate if you share this podcast with other people. I am certain you know people who can benefit from this podcast, and it will take you 10 seconds to share it with them. Just copy the link and then click whatever messaging app you use and send it to the people that you think can benefit from this. They would appreciate it, I will appreciate it, and you will do the right thing as far as letting more people know about what's going on in the AI space and how to learn how to implement it effectively. And if you're there, it will also be awesome if you can give us a five-star review on your favorite podcasting platform and write something that will help us know how and if we're helping you with this podcast. And if you have new ideas on what we should change, I read all of these reviews, and I would love to learn how I can make this better for you. That's it for today. Have an amazing rest of your weekend, and I'll be back on Tuesday.