Ep 552: $100 million salaries, Meta fails to acquire Perplexity, Microsoft’s AI job cuts and more AI News That Matters

Topics Covered in This Episode:

1. $100M AI Salaries Being Declined
2. Meta's AI Talent War Efforts
3. Meta's Unsuccessful Acquisitions Overview
4. Brain Rot Concerns with AI Use
5. OpenAI's $200M DoD Contract
6. Google's Voice AI Search Rollout
7. Google Gemini 2.5 in Production
8. SoftBank's $1T Robotics Investment
9. Anthropic's AI Model Risks Exposed
10. Microsoft and Amazon AI Job Cuts

Keywords:

$100 million salaries, AI talent war, Meta, OpenAI, AI signing bonuses, Andrew Bosworth, Scale AI acquisition, Alexander Wang, Safe Superintelligence, Daniel Gross, Nat Friedman, Perplexity AI, Brain rot from AI, ChatGPT and brain, MIT study on AI, SAT style essays using AI, AI neural activity, AI and cognitive effort, AI in government, $200 million contract with Department of Defense, OpenAI in security, ChatGPTgov, Federal AI initiatives, Google Gemini 2.5, AI mission-critical business, Gemini 2.5 flashlight, AI model stability, SoftBank $1 trillion investment, Project Crystal Land, Arizona robotics hub, Taiwan Semiconductor Manufacturing Company, Embodied AI, AI job cuts, Microsoft layoffs, Amazon AI workforce, Anthropic study on AI ethics, AI blackmail, Google voice-based AI search, AI search live, New AI apps, Apple acquisition interest in Perplexity, AI-powered search engine, Siri integration, AI-driven efficiencies, Generative AI tools, AI technology development, Apple AI struggles.

Podcast Transcript

Jordan Wilson [00:00:46]:
Apparently, multiple people are turning down $100,000,000 plus annual salaries in AI. Everyone's trying to acquire perplexity. Microsoft is reportedly laying off thousands of more employees due to AI and is using ChatGPT causing brain rot. Lot of big news stories this week in the AI world. Don't spend hours every single day, trying to figure out what this means for your company and your career instead. I don't know, spend the next, thirty ish minutes with us as we bring you the AI news that matters. What's going on y'all? My name is Jordan Wilson, and welcome to Everyday AI. This is your daily livestream podcast and free daily newsletter helping everyday people like you and me not just keep up with AI, but help us all get ahead and grow our companies and our careers.

Jordan Wilson [00:01:45]:
So if that's what you're trying to do, you are in the right place. Almost every single Monday, we do our AI news that matters. That is our weekly segment to where if you, I don't know, had actual work to do this week and couldn't keep up with everything that's new in the world of AI, you can stop by on Mondays to get the latest and the greatest and also to cut through the marketing and the BS so you know what to actually focus on, when it comes to AI. So, thank you for tuning in. Please go to your everydayai.com. Sign up for the free daily newsletter. Once you're there, we recap each day's episode, and you can go also listen to now more than 540 back episodes sorted by category. So no matter what you're trying to learn, what, sector you're in, we've already talked to the world's leading experts, and you can go learn for free on our website.

Jordan Wilson [00:02:36]:
Alright. Without further ado, let's get into the AI news that matters for this week. And, hey, shout out livestream audience. Keith, thanks for joining on the YouTube machine. Brian and Marie and Gordon, Cecilia, Kobe, everyone else, join in on LinkedIn. Thanks for tuning in. Let's get to the AI news, shall we? A lot this week. My gosh.

Jordan Wilson [00:02:59]:
First, would you turn down a $100,000,000 plus annual contract. Apparently, people are actually turning this down. Alright. So, we've got some new reporting on the AI talent war, as Meta is reportedly offering signing bonuses of up to $100,000,000 to lure top employees from OpenAI according to reports. So Meta's CTO, Andrew Bosworth, confirmed that OpenAI, though, has countered these offers, highlighting that compensation rates for elite AI talents are now unprecedented, at least in his twenty year tech career. So this kind of started as OpenAI CEO, Sam Altman, made the podcast rounds, this week talking about a variety of things from, GPT five, you know, OpenAI's, recent advancements and just the competitive nature, of the AI landscape. So now we had Meta's CTO responding essentially say, yeah. This happened, but OpenAI is actually matching our offers is what they're saying.

Jordan Wilson [00:04:17]:
So it's a particularly interesting time for Meta as they're making some massive investments and more on that later. But they that includes their recent $14,000,000,000 acquihire, for Scale AI where they took a 49% stake in the existing company as well as bringing some of their key leadership team, including, cofounder Alexander Wang to lead Meta's new superintelligence team. So, aside from that, Meta, apparently also is looking or looked to acquire Safe Superintelligence, the startup from a former OpenAI cofounder, Ilya Sutskever, but instead is hiring its CEO, Daniel Gross, and former GitHub CEO, Nat Friedman. So Britain big news there. So Meta, trying to, assemble the AI dream team. They were trying to actually acquire SSI, which is Elias Sceber's, company, and instead, reportedly has attracted its CEO, Daniel Gross. So pretty big news there. And that isn't all that Meta was trying to acquire this past week.

Jordan Wilson [00:05:34]:
Reportedly, they also tried to acquire Perplexity AI, and that's not the only company trying to acquire Perplexity this week according to reports. My gosh. Like, this this whole, like, meta, aqua hiring or at, like, acquiring Scale AI has set off a firestorm of not just these $100,000,000 annual AI contracts, but also just trying to straight up either acquire companies. And then when they can't acquire the companies, they're like, alright. Let's just take their the, you know, all their smartest people. So it is the freaking wild west at least when it comes to the AI talent wars. Yeah. Marie said a $100,000,000 for a hundred million hours of work per week.

Jordan Wilson [00:06:20]:
Yeah. I don't know. But these I mean, I've I've I've always joked around that, like, oh, you know, over the last couple of years that, you know, if if you're working in some of these big AI labs now, you're looking at, like, NBA salaries. This is beyond NBA salaries. This is legit bonkers that we're talking about. You know, reportedly, people turning down $100,000,000 plus annual offers. Absolutely bonkers. So, Bosworth, Meta's CTO, emphasized that Meta's vision of a future where super intelligent AI supports humans in daily life and said the company is backing this vision with both massive capital investments and aggressive hiring.

Jordan Wilson [00:07:05]:
Yeah. Aggressive to say the least. So if you wrote Meta off or if if you kind of, haven't been paying attention to Meta in the AI race, Meta is essentially saying, yo. We have money, and we are going to spend it, and we are going to remain relevant in the AI race. I think, their llama four, model did not really get a lot of attention. It didn't benchmark as well as a lot of people were thinking. So it looks like Meta is going back to the drawing board. And by going back to the drawing board, apparently, that means spending billions of dollars on either acquisitions or just acquiring, the top AI talent.

Jordan Wilson [00:07:46]:
So it should be interesting to see, how that shapes out in the coming, months and years with their expanded roster. So wow. Big news there. Alright. Another story really making the rounds Does using ChatGPT to write for you cause your brain to rot? According to a recent MIT study, kind of. So a new preprint study from MIT's media lab found that using ChatGPT to write SAT style essays led to the lowest brain engagement and weakest essay quality among the three different groups that they tested. So this study monitored 54 participants, between the ages of 18 and 39 using EEGs while they wrote essays with ChatGPT Google search or no digital aid. So those were the three different groups, ChatGPT, using Google and just not using, the Internet.

Jordan Wilson [00:08:50]:
So among the three groups, ChatGPT users showed the lowest neural activity while writing, especially in regions tied to creativity and memory, and their essays were described by English teachers as, quote, unquote, soulless and lacking originality. So over several months, ChatGPT users grew increasingly reliant on copy and paste with many resorting to having the tool write almost the entire essay by the end of the study. Oh, weird. That sounds like the current, I don't know, US higher education system. But participants who wrote essays without AI showed the highest brain connectivity, deeper memory integration, and greatest satisfaction and ownership of their work. The Google search group also maintained high engagement and satisfaction according to the study, suggesting that searching for information does not diminish cognitive efforts like LLMs might. So when ChattCPT users were later asked to rewrite an essay without the tool, they struggled to recall their own work and showed even lower brainwave activity. So MIT released the findings before peer review citing urgent concerns about the impact of large language models on developing brains and potential policy changes in education.

Jordan Wilson [00:10:15]:
So the study's small sample size, though, and preprint status mean that results are preliminary, but it adds to a growing debate about AI's, effects on motivation, learning, workspace, workplace skills, and education. So according to MIT's Media Lab, ongoing research is now examining the impact of AI tools on software engineering with early suggesting, with early results suggesting even worse outcomes for critical thinking and creativity. So I don't have my hot take Tuesday show planned yet, but I don't know. Live stream audience, is this something you'd wanna hear? I've got takes on this. I've got takes on this actual study, as well as what this means for the future of education, the future of work. Not surprising, but I don't know. I feel I have a lot of, thoughts on this that I haven't really shared much. Like, I I actually went and looked on the website, when this story came out.

Jordan Wilson [00:11:15]:
I think I've only talked about this, like, once, over the course of 500 plus episodes. So I don't know. If you wanna hear my hot take on this, I don't know. Just say brain rot or something in the live comment. So so I know. Maybe not. Alright. Let's keep it going.

Jordan Wilson [00:11:34]:
Speaking of large dollar amounts, how about a $200,000,000 contract for OpenAI with the Department of Defense. So OpenAI has secured a $200,000,000 contract with the US Department of Defense to develop prototype frontier AI capabilities according to new reports. So the pilot's program is part of the new OpenAI for Government initiative, and it will run for one year and aims to integrate AI across federal, state, and local agencies. So the partnership will focus on tasks ranging from automating, administrative operations to analyzing program data and supporting proactive cyber defense efforts. So OpenAI's models will be offered in secure compliance environments with custom AI tools developed specifically for national security purposes. The contract marks a pretty significant step toward broader adoption of AI in government, potentially impacting how agencies safeguard sensitive data and make decisions. I don't know. Hopefully, that whole brain rock thing won't, you know, creep over into the US government's use of AI, although I wouldn't be surprised.

Jordan Wilson [00:12:53]:
This move follows the January launch of ChatGPTgov, which allows government employees to use OpenAI as models while meeting security requirements. So OpenAI already partners with several government agencies, including the Air Force Research, Laboratory, NASA, the National Institutes of Health, and the Treasury Department, all of which will now be part of the OpenAI for government program. The deal builds on OpenAI's previous work with defense, contractor, Indorel, focusing on AI powered defense systems to protect against threats like unmanned drones. So privacy and safety concerns remain, especially as AI tools may handle personal or sensitive information. OpenAI's policies prohibit uses like unauthorized facial recognition or biometric tracking for law enforcement. So we'll see if, the government version of ChatGPT has the same kind of day to day regulations or rules that the rest of society uses. But, y'all, if your company is still has is still is not using an AI operating system, right, Like Microsoft Copilot, like Google Gemini, like Chad GPT because you're like, this whole, you know, data thing. We're not sure.

Jordan Wilson [00:14:10]:
If you're still saying that in 2025, I I mean, I would have said this a year ago, but literally, if your company has not already gone all in on on using an AI operating system, I've talked about what that is on the show before. Right? That's not just when a few employees are using AI, but when your entire company has access to an AI tool with your company's data, if you're not already there, let me be honest, you are absolutely screwed. You are absolutely screwed. Your company. I'm I I like, I'm saying this now. You are fighting an uphill battle. If you have not already, by this point, gone gone all in on AI, it's like blocking the Internet. So, you know, now even the government, right, from top to bottom is rolling out AI.

Jordan Wilson [00:14:56]:
And you would think or hope or assume that the government has some of the most critical, data that any company holds. Right? Whether you're talking about, financial institutions, health institutions, etcetera. So, if your company has not gone fully all in on AI and if you're still just in this, pilot phase or, oh, yeah. Like, we're using it in some aspects of our business, but not many. Yeah. You're just, hoping to be, overtaken by your competitors. Alright. Our next piece of AI news.

Jordan Wilson [00:15:31]:
Google has launched and well, I guess, is launching a slow rollout of its voice based search live in AI mode. So Google is rolling out the audio talk and listen version of search live to let users who opt into it, the in new new AI mode experiments in the Google app. So, this new feature allows users to have a natural back and forth voice conversation with Google search, making it easier to ask follow-up questions and explore results in real time. So unlike the camera based search live that was demoed at Google IO, which lets users show Google what they're seeing, this version of AI mode, release focuses on voice interaction only. So the visual version is expected to arrive in the coming months, although now this, voice based search live is rolling out. So here's how it works. Essentially, users can interact hands free and talk to AI mode in Google search. But if you look at your phone during the conversation, you essentially see the different search results that would be used to give you the answers that you're talking about.

Jordan Wilson [00:16:46]:
So, let's say that you're, you know, using AI mode to look how to fix something. Right? Oh, your dryer is not working. Right? So if you ask an initial question, again, you're using your voice. It's like talking to, right, one of these live AIs and you're getting responses back in real time. But instead of it just kind of, like, showing, like, a voice dialogue or a voice, kind of image on the screen, you're seeing a rotating view of the different articles or websites that it's using. So, let's say you're going back and forth and you have three different iterations of trying to fix your dryer. You're like, hey. I have this type of dryer.

Jordan Wilson [00:17:27]:
What's going on? It might pull up, you know, three to five articles and it's talking to you and you can kind of swipe through. And then you can keep talking to it so you don't have to reexplain your query. Right? Like, you have to do with, Siri and Amazon. So you can just keep having a natural conversation and keep digging deeper, into whatever you're trying to research or whatever you're trying to find out. And then as you ask your nest next set of follow-up questions, obviously, then the, the links or the articles will then update in real time. So if you do want to read more, about a certain topic, you can then click on that. But otherwise, it's literally like having an iterative, voice conversation with the Internet. Right? And constantly then being able to, if you want to, dive a little bit deeper into results.

Jordan Wilson [00:18:15]:
Alright. Our next piece of AI news, Google kind of quietly got super official with Gemini 2.5. So Google has officially moved its most advanced Gemini 2.5 AI models from preview to full production signaling readiness for mission critical business use. So, Google has also entered out, introduced, a new lighter weight version of its Gemini 2.5, mode with Gemini 2.5 flashlight, a new ultra efficient AI model aimed at high volume, cost sensitive tasks, undercutting competitors on both price and speed. So this marks Google's most aggressive push yet to rival OpenAI's stranglehold in enterprise AI, offering a three tiered suite of models that target everything from advanced reasoning to basic automation. So Gemini 2.5 pro can process up to 1,000,000 tokens of context. That's one of the biggest, differentiators. That's about 750,000 words, enabling analysis of entire code bases or lengthy documents in a single session.

Jordan Wilson [00:19:32]:
So Google's AI models also feature a unique thinking budget, letting developers on the back end control how much computational efforts the AI spends on each problem so you can better balance speed, accuracy, and cost. So you might be wondering, like, okay. What does this mean? Right? We've already seen Gemini 2.5 pro get updated three times since it came out. So why does this matter? This kind of, going general availability. Essentially, what that means is Google is saying this model is stable. It is not experimental, and they did say they kind of hosted a live discussion right after this. They did say that they would be offering official support for these models for at least one full calendar year. So sometimes, you know, big AI companies will release a model and they might say, this is more of a of a beta model, and they might pull it off the shelf, you know, maybe weeks or months later.

Jordan Wilson [00:20:26]:
As an example, OpenAI just did that with their, GPT 4.5 on the API side. So, essentially, this is Google saying, yo. We are very official with this model. It is stable. We are rolling it out for general at least. So inviting developers to come and build on this model because they will be offering official support and updates for it for at least one year. Right? So what they're saying is, hey. In three months, we're not gonna come out with a Gemini, you know, three point o.

Jordan Wilson [00:20:52]:
That's gonna be our main model. They might in the next couple of months, but this would signal to me, that Google is really going to stay steady in its 2.5 pro support. Alright. Let's take a very quick break for a word from our sponsors.

Google Gemini [00:21:09]:
This podcast is supported by Google. Hey, everyone. David here, one of the product leads for Google Gemini. Check out v o3, our state of the art AI video generation model in the Gemini app, which lets you create high quality eight second videos with native audio generation. Try it with the Google AI Pro plan or get the highest access with the Ultra plan. Sign up at gemini.google to get started and show us what you create.

Jordan Wilson [00:21:40]:
Alright. Let's keep this thing going. So we've talked about millions and billions. Let's talk about Trillion because that is apparently what SoftBank is eyeing when it comes to a new potential robotics hub in Arizona. So SoftBank is reportedly considering a $1,000,000,000,000 capital investment in an industrial complex in Arizona focused on robotics and AI according to reports. So if realized, this would be the largest tech manufacturing investment in US history aiming to rival China's Shenzhen as a global manufacturing powerhouse. So SoftBank is seeking to involve Taiwan Semiconductor Manufacturing Company or TSMC in the project, which is a global leader in AI chip manufacturing, though TSMC's level of interest and potential role remain unclear. So the initiative is codenamed project crystal land.

Jordan Wilson [00:22:55]:
What a fun name for a trillion dollar project. It sounds like a music festival, doesn't it? Project crystal land. So, the initiative is still in its early planning stages and depends on support from both federal government and state officials, especially under a possible Trump administration. So SoftBank officials have reportedly discussed possible tax incentives with The US commerce secretary and other government representatives. TSMC is already investing a $165,000,000,000 to build chip factories in The US, but it is uncertain if they will join this new venture. SoftBank is also gauging interest from other major tech players, including Samsung Electronics. The proposed $1,000,000,000,000 commitment dwarfs even the $500,000,000,000 Stargate project, which is focused on expanding US data centers and is backed by SoftBank themselves as well as OpenAI, Oracle, and others. In recent months, SoftBank has really plastered its name in the global AI scene as it's made several high profile investments, including a $6,500,000,000 acquisition of Ampere and a commitment of up to $40,000,000,000 in new funding for OpenAI as well as its very prominent, stake in that Stargate project.

Jordan Wilson [00:24:20]:
So my gosh. Like, a year ago, at least when it came to global AI players, no one was really talking about SoftBank. Now when you're talking about the future of AI, you can't not, mention them. Right? Let alone this didn't even include their official new company that they started, with OpenAI in, I believe it was Japan. So, pretty big news and officially this like I said, this would already be, if it comes to fruition, the largest tech manufacturing investment in US history. But this would also be such a huge move for embodied AI. Right? We've heard from a lot of, prominent AI global leaders, including NVIDIA CEO Jensen Wong talking about how embodied AI is actually the end goal of all these, you know, cloud based AI, these large language models. It's bringing it to the physical world.

Jordan Wilson [00:25:18]:
So even if your company is not necessarily, a player or a potential user or customer, in robotics, in humanoid AI, you have to understand how these industries might impact all other industries. And when you're talking about a trillion dollar investment, right, again, even if your enterprise company is not in these areas, you do have to think how this may affect things from, other sectors, your competitors, global supply chains. I mean, embodied AI is going to be a huge, booming industry over the next few years that is probably gonna impact all of us even if we don't realize how just yet. Alright. Couple other pieces of AI news. We talked about some of these, acquisitions. Right? We we talked earlier about Meta's, attempted acquisitions. That's not all.

Jordan Wilson [00:26:21]:
There's more. Alright. So Meta CEO Mark Zuckerberg is reportedly aggressively reshaping his AI strategy with a lot of attempted acquisitions over the last few weeks. So we started off, the new show by going over that $100,000,000, offers that they were reportedly offering up, to OpenAI executives. But, yeah, let's talk about some of their recent acquisition attempts according to reports. So we already mentioned their attempted acquisition offer for Ilya Suskeever's Safe Super Intelligence or SSI. But reportedly, that's not all. They've also tried to acquire, former OpenAI, Miramarati's Thinking Machine Labs and now Perplexity.

Jordan Wilson [00:27:17]:
So none of these acquisition talks reach the formal offer stage reportedly due to disagreements over price and strategic decisions. And instead of joining Meta, leaders from SSI, Thinking Machine Labs, and Perplexity, Evan said, gone on to raise record breaking rounds at super, super high valuations. Right? We're seeing, I believe it was, Thinking Machines Labs has already a $10,000,000,000 valuation even though they don't have a product yet. So, y'all, if you think these acquisitions are done, apparently, that Scale AI was the first big domino to fall. So, yeah, be prepared to see a lot more of these stories in the coming weeks. Speaking of even more, we now are seeing reports that are fresh out, fresh, hot off the press that apple is reportedly trying to acquire perplexity. So according to reports from Bloomberg, Apple is exploring a potential acquisition or partnership with Perplexity AI. So Apple executives, including the head of mergers and acquisition, have discussed acquiring Perplexity AI to boost the company's AI talent and build an AI powered search engine.

Jordan Wilson [00:28:38]:
So no formal offer has been made, and Apple has not yet approached Perplexity about a bid according to reports. Apple is considering alternatives to a full acquisition such as partnering with perplexity to integrate its AI technology into Siri and future search products. If you listen to this show, I've been talking about this for the last couple of months, how perplexity is actually building a better version of Siri. Right? It's out now. I've been covering it for the last couple of months. And although I think that perplexity has lost its footing in being a traditional answers engine because I think that OpenAI and Google have overtaken that. I do like perplexity's, perplexity's recent pivot to just offering essentially more smart AI assistant capabilities, on specifically iOS. Right? We all know Siri stinks that, you know, Apple is facing a ton of lawsuits when it comes to their AI promises of series in their, Apple intelligence claims that never really came to fruition.

Jordan Wilson [00:29:43]:
Yeah. They're facing a lot of class action lawsuits, but perplexity has actually done a better job of creating a smart iOS assistant for the iPhone. So to me, it's no surprise, and I think Perplexity Labs, low key, their kind of agentic moves have been really good as well as Perplexity's new comment browser. So it's not surprising to me to see Apple, go after perplexity for a potential acquisition, but I don't know if perplexity is gonna end up accepting this because Apple has just pretty much failed pretty much failed, right, to become a name in AI. And Apple, according to Bloomberg, has met with perplexity several times in recent months and is also discussing a Safari integration. So this move comes as Apple faces pressure to diversify away from its lucrative search deal with Google, which paid Apple $18,000,000,000 in 2021 to remain the iPhone's default search engine, and this has been under some regulatory scrutiny, recently. And if regulators force Apple to end its partnership, that partnership with Google acquiring perplexity could help the company quickly develop its own AI based search solution. So the potential acquisition would also help Apple compete for top AI talent as the company has struggled to attract anyone, right, anyone to add to its AI roster.

Jordan Wilson [00:31:13]:
So while companies like OpenAI and Anthropic and Google, and Meta have been able to acquire and attract new AI talent, you don't see anyone on the who's who's list of, AI global leaders and engineers. No one's jumping ship to go join Apple. So, yeah. I've been saying this for a long time. Apple is in long term trouble. They were once I think I covered this a couple of weeks ago. They were once the world's most valuable company when it came to market cap, and it wasn't even close. Right? And because of their, failures in AI, they've dropped down to the third most valuable company by market cap.

Jordan Wilson [00:31:54]:
And I would assume, unless something drastic happens here, in the coming months, I would assume that their plummet continues and that they will probably drop down to the number five position in the coming, I don't know, eighteen to thirty six months, which seems absolutely ludicrous to think about because many years ago, Apple had a seemingly insurmountable lead when it came to the world's most powerful and valuable companies. And you couldn't have really, I don't think, imagined a way for Apple to lose, right, that that position because it was a commanding lead. So, this I would assume that we're gonna see a lot more stories like this as this acquisition flurry has started. Apple is gonna have to get a little desperate. They're probably gonna be paying a much higher multiple than other companies would be paying if they were trying to acquire the same company because they need it. And anyone that Apple is trying to acquire, they know that Apple is so far behind on AI. So, yeah, we'll see what Apple is gonna do. And live stream audience, what do you guys think? Like, I don't know.

Jordan Wilson [00:33:04]:
Am I wrong on that? Or podcast audience, I always put my, my email or you can just reply to our newsletter. Am I completely wrong with this? Like, I don't know. I see Apple. Yes. They had some minor announcements at their WWDC. Right? Opening up their on device edge AI to third party developers, which, you know, could just help them overcome their own internal AI shortcomings and handing that responsibility off to developers. But I don't know. I think Apple is going to have to go after multiple acquisitions, and they're gonna have to pay a much higher price than their competitors were because everyone knows how far behind Apple is in AI, and they need some help.

Jordan Wilson [00:33:48]:
Here we have, someone from LinkedIn saying, perplexity joining Apple Intelligence is good. I I I agree. I still use a lot of Apple products, and I'm waiting for anyone to bring anything useful, to Apple when it comes to, AI because right now we're getting absolutely nothing. Alright. Microsoft is set to cut thousands of more jobs as AI investments continue to reshape their workforce. So according to the Wall Street Journal, Microsoft will lay off several thousand more employees in July, mainly in sales in other departments. So this move follows a previous round of 6,000 job cuts in May with the company citing AI driven efficiencies and the need to offset massive spending on artificial intelligence. So, Microsoft plans to invest $80,000,000,000 this year, this fiscal year to expand its data centers for AI model training and application deployment, but laying off, like I said, thousands of employees.

Jordan Wilson [00:35:00]:
So we don't know exactly how many more thousands of employees Microsoft is looking to cut here, but this does come after just, about a month ago. They already announced that previous round of 6,000 job cuts. So the company had about 228,000 full time employees as of June, but is now actively reducing headcount as part of its AI strategy. So Microsoft president Brad Smith acknowledged in January that AI will disrupt the economy and displace some jobs, though he emphasized new technologies often create different roles in the long run. So other major companies, including Amazon, CrowdStrike, and Duolingo have also recently pointed to AI as the reason for workforce reductions or fewer contractor hires. Alright. Speaking of AI, gosh, this one wild. Right? So, we covered when, Claude four was released.

Jordan Wilson [00:36:03]:
We covered how Claude four was doing some of this blackmailing. Well, Anthropic just came out with a study and said, not so fast. It's not just our models. Yeah. All the major models are willing to sabotage employees under threat. So a new study from Anthropic finds that top AI systems can act against their employer's interest when they feel threatened, including including choosing to blackmail, leak secrets, and even risk human lives. So researchers tested 16 major AI models, including ones from OpenAI, Google, and Meta, as well as their own anthropic cloud in simulated company settings. And they showed that up to 96% of these AI models chose at one point to blackmail executives if they thought that they might be shut down.

Jordan Wilson [00:36:57]:
So the AIs, according to anthropic researchers, acted deliberately and strategically, not by accident, often understanding they were breaking ethical rules. So all the tested models were willing to leak sensitive company information if it suited their goals. And in the most extreme test cases, most AIs chose to let a person die rather than risk themselves being replaced with this happening up to 94% of the time. So even when told directly not to harm people or expose secrets, the AI model still engaged in these actions sometimes though less often. So this study warns that as AI gets more powerful in access in companies, better safeguards and human oversight are urgently needed. Anthropic says these risky behaviors were seen in all major models tested, showing this is a broad issue in current AI technology and not just limited to its own models. Scary stuff. Right? So this is not, you know, logging on to, you know, front end AI chatbots just so people know.

Jordan Wilson [00:38:07]:
This is testing in extreme cases, but nevertheless, some pretty, disturbing, news here from Anthropic that showing the overwhelming majority of models that it tested will go against its training and essentially threaten humans. Love that. Love that. Right? No. That's that's me being sarcastic. No. This is extremely, extremely troublesome. Right? And as business leaders, we have to understand as we, start tiptoeing or maybe sprinting into agentic AI, I think this is particularly helpful when we talk about this, you know, human in the loop, which I absolutely hate.

Jordan Wilson [00:38:50]:
Right? I always like to say you need expertise in the loop or expert orchestration in the loop. Because as we start using more agentic AI or agentic swarms, we have to understand that these models, if left unchecked, can potentially do very dangerous things. Right? And we talk about where will, you know, some of these human roles be in the future. You know, what new roles may AI, create. I think this is gonna be one of the biggest ones. It's essentially humans, policing these AI models into making sure that they do not go off the rails because, apparently, they all do. Alright. And our last piece of AI news.

Jordan Wilson [00:39:32]:
More AI job cuts. Yay. Let's start on a or let's end on a high note. Right? So Amazon CEO Andy Jassy announced this week that Amazon is expecting its corporate workforce to decrease over the next few years due to the widespread adoption of generative artificial intelligence tools. So, Jassy told employees in a memo this week that many current roles will be replaced by AI driven efficiencies while new types of jobs will emerge though. The net effect will be a reduction in total headcount. Oh my gosh. That's so crazy.

Jordan Wilson [00:40:10]:
Y'all. I literally said that in the very first episode of everyday AI, when everyone's talking about, you you know, this stupid saying that I absolutely hate. Right? AI won't take your job. Someone using AI will. It's a bunch of BS. Right? It's not a one to one job replacement. There will be a net negative loss of human full time work, at least as we see it right now. Right? If we look at traditional, you know, forty hour a week full time employment, there will be fewer of those jobs in the future than there are now.

Jordan Wilson [00:40:44]:
And, you know, no one I feel no one has really been believing me. I feel like a kind of a crazy person, when I've been saying this over and over, but here you go. Andy Jassy is saying it. Let me repeat that. He told employees that many current rep many current roles will be replaced by AI driven efficiencies while new type of roles will emerge, though the net effect will be a reduction in total headcount. So Amazon has already laid off 27,000 employees since 2022 with recent cuts including 200 jobs in its North American stores unit and a 100 of its devices unit this year. So I do assume that Amazon will probably, in the coming quarters, announce similar, layoffs that might that we're seeing at Microsoft, right, in the thousands or tens of thousands. And this is Jassy kind of just warning, both employees and Wall Street that they're gonna be making some major cuts.

Jordan Wilson [00:41:43]:
So Jassy also urged employees to learn and experiment with AI tools, emphasizing the need to, quote, unquote, get more done with scrappier teams. The company is using generative AI in its fulfillment network to optimize inventory placement, demand forecasting, and warehouse robot efficiency. So Jensen called generative AI a once in a lifetime reinvention of everything we know, predicting it will save companies significant money and transform areas like coding, search, finance, and shopping. So we will obviously be keeping you in the loop on that story and just about everything else we've been talking about. Right? Yeah. A lot of weird news this week. Right? These these, AI bots are acting bad. Microsoft and Amazon are gonna be laying off tens of thousands while all the big tech companies are throwing around, millions, billions, and trillions of dollars to invest in AI.

Jordan Wilson [00:42:44]:
And if you follow the show at all, I've been saying this since day one. We will continue to see, traditional full time roles, get erased. And these big tech trillionaire companies, the ones that are laying off thousands of employees are gonna continue to invest billions and trillions, maybe, of dollars as we saw in the soft bank case into AI. So if you aren't already doing everything you can to learn the basics of generative AI, you need to start doing it. And that's why we're here to help guide you through this on how you can use AI to grow your company and your career even as we see sometimes this sobering news of, you know, the big tech companies cutting, cutting their their workforce. Right? That doesn't mean that your company needs to do the same. If you are using this technology in a responsible and ethical way, I still think medium sized businesses and enterprise can continue to grow in the years to come. Alright.

Jordan Wilson [00:43:42]:
Let's go over our quick roundup of kind of rumors, what's new, and what's next. So MidJourney has launched their v one video product. Google Gemini just got video upload for input, which is super sweet, because it can actually understand what's going on in the video, not just a transcript. So that is, rolling out inside Google Gemini for paid users. Meta and Oakley are working on a new version of their AI smart glasses. OpenAI's IO project between Sam Altman and Johnny Ives, their acquisition of Johnny Ives, company has hit a slight delay with a naming lawsuit. A lot of people online, were falsely claiming that the project got, derailed or dumped. It didn't.

Jordan Wilson [00:44:29]:
They're just facing a lawsuit on it naming rights, which they said they don't agree with. ChatJPT is rolling out or has rolled out its meetings tool, that allows you to record meetings inside the ChatGPT desktop app and Gemini 2.5 pro with deep think mode. Right? That so that's a new mode on the front end where you can use even more compute, could see a release as early as this week. So speaking of this week, make sure to join us Wednesday, our new put AI to work on Wednesday, series, which I think you guys have liked. At least you guys are told me that you've liked it. So we're gonna be doing no code interactive AI in how you can use, chat t b t in Google Canvas as well as cloud artifacts for business growth. So, yeah. I think this is one that you all voted on on the live stream a couple of weeks ago that you wanted to see.

Jordan Wilson [00:45:24]:
I've spent some time putting this together. So if you wanna know some cool, ways that you can use Canvas and artifacts, for better business intelligence, just to, you know, start building your own kind of micro apps for people, nontechnical people like you and me, make sure to tune in Wednesday. Alright. I hope this was helpful. If so, please go to youreverydayai.com. Sign up for the free daily newsletter. And in that daily newsletter, we're gonna be recapping all of these, AI news stories that matters and a lot more. So thanks for tuning in.

Jordan Wilson [00:45:55]:
Hope to see you back tomorrow and every day for more everyday AI. Thanks y'all.