Everyday AI Podcast – An AI and ChatGPT Podcast - Ep 720: China Stealing AI from the U.S.? Inside Anthropic's Bombshell Allegations

Episode Date: February 24, 2026

Did China steal Anthropic’s AI powers? Well, that’s the shocking bombshell report that Anthropic just dropped. They accused multiple Chinese AI companies of generating more than 16 million exchanges with their models just to try and copy it. We get what you’re thinking….. “So. That means cheaper open Chinese models so we all win, right.” Wrong. On this episode of Everyday AI, we break down Anthropic’s shocking AI distillation accusations against Chinese firms, what they actually mean, and how they’re more impactful outside of just the AI model you choose to use. You might be shocked TBH at the far-reaching implications. China Stealing AI from the U.S.? Inside Anthropic's Bombshell Allegations — An Everyday AI Chat with Jordan WilsonNewsletter: Sign up for our free daily newsletterMore on this Episode: Episode PageJoin the discussion on LinkedIn: Thoughts on this? Join the convo on LinkedIn and connect with other AI leaders.Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineupWebsite: YourEverydayAI.comEmail The Show: [e-mail] with Jordan on LinkedInTopics Covered in This Episode:Anthropic Accuses Chinese AI Labs of DistillationDetails on 16 Million Claude Extraction PromptsDeepSeek, Moonshot, MiniMax Named in Anthropic ReportGoogle and OpenAI Cite Similar China AI ThreatsTechnical Explanation of Model Distillation AttacksMarket Impact: MiniMax Surpasses Anthropic in TokensFinancial Consequences for U.S. AI Model ProvidersPolicy and Geopolitical AI Competition AnalysisLimitations of Current Export Controls and SafeguardsU.S. AI Dominance Threatened by Chinese DistillationTimestamps:00:00 "Foreign AI Impact on Tech"04:43 "AI Distillation and Security Threats"07:11 "MiniMax Scandal: Data Theft Allegations"10:33 "Open Router Key Marketplace"15:54 "Smart, Cheap Models Explained"19:31 "AI, IP Theft, and China's Impact"22:44 Big Tech's Data Theft Problem25:37 "Protecting U.S. Tech from Export"27:53 OpenAI Accuses DeepSeq of Misuse33:58 "AI's Global Power Struggle"35:03 "AI Models: What's Next?"Keywords: China ASend Everyday AI and Jordan a text message. (We can't reply back unless you leave contact info) Start Here ▶️Not sure where to start when it comes to AI? Start with our Start Here Series. You can listen to the first drop -- Episode 691 -- or get free access to our Inner Cricle community and all episodes: StartHereSeries.com Also, here's a link to the entire series on a Spotify playlist. 

This is the Everyday AI Show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business, and everyday life. Meet Firefly AI Assistant, now live in Adobe Firefly, the All In One Creative AI Studio. Just describe what you want to create and the assistant handles the rest, orchestrating multi-step workflows across Photoshop, Premiere Express, and more in one conversational interface. You direct the outcome. The assistant accelerates execution. Most people think AI competition is a Silicon Valley issue.

It's not. It's an everything problem and an everyone issue. And a breaking report from Anthropic might be the biggest wake-up call we've gotten so far on how AI competition actually impacts this all and not just those living in Silicon Valley. That's because Anthropic just. publicly accused three Chinese AI labs of extracting Claude's most advanced capabilities through 16 million prompts in more than 24,000 fake accounts. Anthropics not alone and the reports are escalating.

I mean, OpenAI just sent a memo to Congress earlier this month telling a similar story. Google reportedly identified over 100,000 distillation prompts targeting their Gemini. And all of this has been in like a. two-ish week span. Probably not coincidentally, Chinese AI models have legit taken off this year, both in terms of capabilities and in usage. So why does it matter? Maybe you don't work at an AI company?

Well, here's why you should care even if you've even never touched an AI tool, because it's about more than OpenAI Anthropic and Google. It's also about Microsoft and meta and Amazon and. a lot of the other big tech companies that maybe just hold a big position in your retirement portfolio. These are the models behind the software your company uses, the platforms your retirement fund is invested in, and the systems that are literally reshaping the American job market right now. So if a foreign competitor can reportedly copy and paste all of that U.S. innovation overnight for pennies on the dollar, the ripple effect.

hits everyone, not just big tech. So we're going to tackle that bombshell report and a lot more on today's everyday AI. What's going on, y'all? Welcome. My name's Jordan Wilson, and I'm the host of this thing. And it's for you. It's an unedited, unscripted daily, live stream, podcasts, and free daily newsletter, helping everyday business leaders like you and me keep up with these nonstop AI developments

because it's quite literally nonstop. And I tell you what to focus on to grow your company and your career. So if that's what you're trying to do, if you're like, hey, I don't have 10 hours a day to take care of this. Do it for me, Jordan. I got you. Go to our website, your everyday AI.com. There, we're going to, you can go sign up for our free daily newsletter. Each day, we recap the highlights from that day's podcast as well as keeping you out to date with all the

other AI news that you need to know. Speaking of things you need to know, yeah, dropping it again, 712, 713. Go listen if you haven't already. That's our 2026 AI prediction and roadmap series. So is China really stealing AI from the U.S.? Anthropic says so, and they've joined OpenAI as the latest to accuse China of distillation. And this is a pretty shocking report because we've heard this before. We've heard it from Open AI.

We've heard it from Google and we've heard other big tech companies kind of mumble about it, right? But no one's directly coming with like receipts. and straight up pointing fingers. I think Open AI has probably been the closest. They even went to Congress about that, but we're going to unwrap that here in a bit. But this is a pretty big story. So on today's show,

we're going to unpack Anthropics bombshell accusations against these three Chinese AI labs. We're going to explain what the heck distillation even is. If you're a non-technical person, I'm going to tell you, well, why every major U.S. lab is sounding the alarm right now and expose the elephant in the room that makes this story far more,

complicated and also interesting. All right. So here's what it started with. Yesterday, Anthropic put out essentially a blog post and obviously it's going to make its round in the media today and probably throughout the rest of the week. But here's what they said. I'm just going to read a little snippet or the beginning of their post and their announcement so you don't have to.

So they said, we have identified industrial scale campaigns by three AI laboratories, Deepseek, Moonshot, and Minimax to elicitly extract Claude's capabilities to improve their own models. These labs generated over 16 million exchanges with Claude through approximately 24,000 fraudulent accounts in violation of our terms of service and regional access restrictions. These labs use a technique called distillation, which involves training a less capable model on the outputs of a stronger one. Distillation is widely used in a legitimate. training method. For example, Frontier AI Labs routinely distilled their own models to create smaller, cheaper versions for their customers. But distillation can also be used for illicit purposes.

Competitors can use it to acquire powerful capabilities from other labs in a fraction of the time and at a fraction of the cost that it would take to develop them independently. These campaigns are growing in intensity and sophistication. The window to act is narrow and the threat extends beyond any single company or region. Addressing it will require rapid coordinated action among industry players, policymakers, and the global AI community. All right. So before I go any further,

I'm going to state the obvious here, right? I'm from Chicago. I live in the United States. Yes, we have a global audience on the podcast. And I'm sure, you know, I'm going to get plenty of, you know, emails or messages about how I'm being overly American or something like this, right? I'm speaking for our main audience, right?

Our biggest audience is in the U.S., so I am taking this through the U.S.'s perspective. So if you don't necessarily care, you know, feel free to hit skip on this one. But I do think it's important because I actually think that this story right here, as it unravels a little bit more as we start to peel layers off the onion, we are going to see it is full. more consequential than just Chinese AI labs reportedly ripping off US labs. It's much bigger than that. So anyways, let's get a little bit more into the details of Anthropics report. So they did accuse DeepSeek Moonshot AI and Minimax in coordinated extraction campaigns distilling their models. And they traced the activity to those specific lab researchers using IP and metadata.

So here's why this matters. And this is according to Anthropic. So they detected minimax before they released the model being trained. All right. So essentially, if you haven't heard minimax, they've legit blown up. And I'm going to bring some of my own receipts here in a little bit, why I think that they don't singled out minimax necessarily, but why I think minimax is especially important in the broader scope of this story. So Anthropic said that Minimax redirected nearly half of its traffic to capture the newest systems capability. And one proxy network managed over 20,000 fake accounts simultaneously to avoid detection. So this isn't, you know, again, reportedly, right?

This isn't just a couple rogue researchers at a lap, right? This is a concentrated effort nationally in China for their biggest labs to, according to the labs themselves, steal their data, right? And train their models on their respective work. So now let's get into a little bit of the receipt. So I have mentioned this once or twice over the past three weeks and, you know, talking a little bit more about the anthropic. and Open AI and Google kind of competition here domestically. But you have to rewind a little bit to really understand. So without getting into too much detail, Anthropics bread and butter is developers.

It's the API, right? Anthropic doesn't have, you know, 900 million weekly active users like OpenAI. They don't have, you know, hundreds of millions, you know, of users like Google. Their bread and butter, and according to reports, the majority, overwhelming majority of their revenue comes from the API. So that's when developers, builders, you know, software engineers are using a certain API to accomplish a task. A lot of times it's maybe writing software, building apps, running, you know, financial transactions, right? Everything can be done through code. People think it's this software engineering. No, it's not. Computer.

use, agents, you know, running complex, you know, data queries in certain, you know, financial modeling programs, et cetera, right? So, enthropics, bread and butter is on the API. So there's pros and cons to that, right? The pro is, well, the price is not capped. So whether you're paying $20 or $200 a month for Chad ChbT or Gemini or if you're on an enterprise plan, right, the spend or your cost as a company. is capped. On the API side, it's not. Right. So, you know, maybe let's just say one company might spend,

you know, a million dollars with OpenAI for a, you know, I don't know, a certain number of enterprise licenses. They might spend five or $10 million on the API side with Anthropic, because if they find Anthropics models are good enough to account for that cost, they'll do it, right? So there's no cap to that. So if you go back a year ago, and this is March 2025, if you're looking at open router. So open router is essentially the most popular third party API tool. So you sign up for an open router key. And then it makes it very easy. Actually, if you ever want to switch providers, you can do it without much friction. Right. So it's essentially think of it as a model agnostic marketplace. You sign up for it. And at any time, you can switch things out.

It doesn't mean you shouldn't re-engineer things when you switch things out with a couple of clicks of the button, but it makes it very easy for companies to experiment with different models, you know, but they are the biggest when it comes to running inference, right? So they have the most data in terms of who's using tokens globally. In a year ago, it was Anthropic. Anthropic was in a commanding lead. Again, this ties to Anthropics revenue. Anthropic is reportedly trying to go public any quarter now, right?

We've heard by the end of 2026. And their month over month revenue is reportedly stalling. Still growing at a crazy rate, just not growing at the same rate it was. So technically stalling is less of a hockey stick and more of just a nice uphill climb. But this has to be a concern because last year, 40% market share on OpenRouter tracking total token share by model author. All right. Last week, the last full week of February, that Open Router tracked 12%. All right. And they are in third place. And guess who is in first place now? Minimax out of nowhere.

Right. So you actually have in the top seven spots. All right. I'll just read them. Number one, minimax now at 20%. Number two, Google. Number three, Anthropic. Number four, open AI. Number five, ZAI. Number six, moonshot, number seven, deep seek. So the three companies essentially that Anthropic accused of distilling their models are now leading the token. share, right? So people might be saying, oh, but Jordan, they're open source. Why does it matter? Don't these, don't companies just download these models and run them locally? Sure, they do, right? But obviously, a lot of companies still, you know, need this speed and the capabilities of using these models in the cloud. And they're obviously, right, using them in the cloud. Because right now, if we look at Minimax, mini max has almost three million or sorry three trillion tokens that they used last week via open

router. This is just via open router, right? So the actual number is a lot more. So I want you to keep that in mind. Let's do a little math here, if you may, or if we may, right? I'm not going to make you do the math. We can try to do it together. So Anthropic went from a year ago, a 40% share down to a 12% share now in terms of total tokens used in open router. All right. So let's just use that three trillion versus where Anthropics at right now, which was about 600 billion. So much less. So if we use a three to one mixed ratio, all right, of token usage, because we don't know, you know, cashing, input, output, we don't. Let's just ballpark some numbers here. Right. So let's say 75% input, 25% output, which is pretty standard.

if anframpic was still in the top spot where minimax is now not even saying if they had the same commanding lead they did a year ago right even if their lead shrunk and if they were just in minimax minimax's spot right that's about roughly 13 to 22 million dollars a week of lost revenue all right it's not a ton but on the high side that's a billion dollars right and that's just one this is just through open router so i'm sure that there's plenty of losses right people directly working uh obviously in the uh claude council who are no longer or maybe reducing their spend there so my hunch my hunch um is anthropics size of the token pie even though the pie is growing their slices so much smaller and my guess is

the fact that Anthropic came out guns blazing because no one's done this, right? Open AI surprisingly did it very suit and tie, right? They didn't come out with these kind of receipts, you know, against the companies. Maybe they didn't have them. Maybe they were saving them, you know, in case they needed to do this in some, you know, litigation type setting, although we don't know if that will be possible because it's China. But still, you're looking at. at probably even just in open router, right, for what we can track, probably 750 million to

a billion dollars a year. And it's probably, like I said, multiple billions of year as other developers are now saying, wait, if I can get mini Macs, right? So here on my screen for our live stream audience, I'm showing the artificial analysis, kind of the cost to run intelligence index, right? So this is, you know, on the different axes, you have axes, axes, axes, right? I learned this 20 years ago, right? I can speak. But you have your intelligence index. So smarter models, right, are measured vertically, right? So you want to be on the top. And then the models that cost less to run are on the left. So the ideal quadrant is the super smart models that are somehow super cheap to run, right? Which is very, very rare.

And you'll see of the models in that quadrant, it's basically all China and Gemini 3 Flash, right? So you have Minimax, Deep Seek, Mimo, Kimmy, and then Google Gemini 3 Flash. So what does this mean? And I've been saying this, right? So ever since the 2025 Deep Seek debacle, and I told you guys, don't fall for it. I literally said, this is not real. Don't fall for it. And I said more stuff's going to come out.

But deep seek is clearly, you know, not doing this all themselves. I'll just say that because I don't want to write, write any checks. I can't cash with my mouth, right? But Open AI has essentially said that, yeah, Deep Sea's doing this. And now Anthropics said it as well. So I think it's safe enough to say, right? The consensus is, well, Chinese companies are distilling their models. The models are getting better.

and they're getting cheaper because presumably the Chinese companies don't have to pay as much. They're paying pennies on the dollar. So let's understand the basics of model distillation, right? So for our non-technical audience. So think of it as student teacher. So there's a student model that learns by studying a teacher's responses, right? So essentially the teacher's like, yo, here's the test, here's the answers, and here's my reasoning and rationale behind this.

Right. So Frontier Labs use this to legitimate. in-house create cheaper versions of their own model, right? And we've even seen and heard that companies are even using their own models to, you know, create versions of themselves, you know, self-replicating, right? All that crazy stuff. That's not what we're talking about here. So think in most of the big frontier labs have talked about distilling smaller versions of

their model based on the big, powerful model, right? But the controversy starts when competitors, to it. So distillation in and of itself is a very normal thing to do within company A, right? Company A, distilling their own model to create model A, B, normal. Company C, distilling from company A, not normal. That's authorization at a massive scale, right? I was thinking of a good example, right? This would kind of be like if there was a new streaming, video streaming service startup that just decided to record every single Netflix original. And then they went in there and they just kind of use some CGI to change a few scenes,

you know, change the faces on there. And then they just repackage the content under their own brand. And then they launched a competing platform overnight, right? Probably doing it at 1% of the cost and 1% of the time. That's what model distillation is. And right now, well, there's no international, obviously, law against this. And there's no domestic law against this. and number three, even if there were, there's really no way necessarily for U.S. companies to

hold Chinese companies accountable. And I think, you know, without saying too much, I think we can all understand how, you know, China's reputation in this realm kind of might and probably will and reportedly has bled over into the AI space. And that's not just me mumbling some range. FDem facts, right? So when it comes to IP theft, right, which is a different but kind of related issue, right? The FBI and the U.S. Trade Representative estimate the cost of IP theft from China is between $225 billion to $600 billion each year. And some experts estimate right now that over

the last two decades, that cumulative impact has resulted in a wealth transfer of four to $12 trillion away from the U.S. economy. All right. So if AI is the new oil, if AI is the new military superpower, if AI is the new currency, then you can probably understand how China's repeated IP theft, again, according to the FBI and the U.S. Trade representative, between $225 billion to $600 billion. per year, you can probably understand this whole AI race. It's happening for a reason.

It's not just to say, oh, you know, we have some cool new things that we can do at our company that we didn't do before. No, I've said before. AI is more important. Excuse me there. I'd like I do think AI is more important than oil. It's more important than gold. It's more important than natural resources. It's more important than on like I hate saying this. It's more important than probably anything right now, at least for the future. Maybe not to day, but for the future. That's why you have every single company around the world putting every dollar they can. Right. I think a story in our newsletter today, it was more than $600 billion. U.S. just U.S. big tech companies are investing into AI infrastructure this year, right?

Every single company around the world wants to have the best models. In China, reportedly, They're fine with just kind of taking the work from others and repackaging it as their own. So this is how distillation becomes a weapon. So labs are using kind of these proxy services and thousands of fakes accounts to access frontier models, mainly from the U.S. And it is targeted prompts that extract these high value capabilities like the reasoning. So it's not just the output. It's also the reasoning and how a model got to it. Because, yeah, you can show, you know, if you look at the teacher-student model there,

The teacher can give you the answers, but if they don't show their work, maybe the student can't learn as quickly. So that's why I think that this has gone even faster and further as reasoning models have been out for a little bit longer. And some attacks actually force models to reveal step-by-step reasoning for richer training data. Adobe just introduced an entirely new way to create, bringing the power and precision of its creative suite into one, conversational experience.

Meet Firefly AI Assistant, now live in the Adobe Firefly app, the all-in-one creative AI studio. Powered by Adobe's creative agent, Firefly AI Assistant lets you start with your vision, just describe what you want, and shape the outcome as it takes form with the Assistant. The Assistant orchestrates multi-step workflows, drawing on 60-plus pro-grade tools across Adobe Creative Cloud apps, including Photoshop, Illustrator, Premier, Lightroom Express, and more to help bring your ideas to life. You can also get started with creative skills, a growing library of pre-built workflows for

common creative tasks, like batch editing photos, creating mood boards, portrait retouching, and creating social variations. Every step the assistant takes is visible so you can refine, redirect, or take over at any time. You stay in the driver's seat as the creative director. Adobe Firefly AI assistant now in public beta. See it today at Firefly. dot adobe.com.

And I get what you're probably thinking, right? Maybe I should have started with this or mentioned it earlier in the show. You're probably thinking, oh, that's rich, right? Stealing data. We can't feel bad for big tech stealing data, right? That's maybe how they got here, right? You can look at any of the big AI labs and you could probably make a legitimate case that at some point, they maybe use some data that wasn't theirs, right?

So Anthropic, I think they're one of the few so far that has actually paid a hefty and noteworthy fine for this. So in September of last year, Anthropic settled a $1.5 billion kind of judgment there in the Barts versus Anthropic class action lawsuit. And the lawsuit alleged that Anthropic use about 500,000 pirated books from shadow libraries like Libgen to, train its clawed model. So, and obviously, you can probably make the argument that other AI labs are facing similar lawsuits like this every single month of varying degrees, right? There's obviously the big one, you know, Open AI versus New York Times. It's been going on for a couple of years. So I'll address the elephant in the room, right? You can say, oh, it's very rich that these AI labs are, you know, accusing someone else of unauthorized data use when maybe that's how they got

to the point where they're at to date. I get that. But I think that we can talk about A, without it being, you know, turning into a B conversation, right? We don't have to have what aboutism, you know, happening in our AI discussions, right? This can be a huge problem. And also unauthorized data training by the AI labs can also be a huge problem. And I think that we have to be able to separate. it because the impacts are so much more than just the AI.

And here's why. Actually, no. First, a little bit more. Because right now, Anthropic is arguing that these Chinese gains are wrongful and seen as proof that export controls have failed. Right. So it's not like U.S. labs have found themselves in this position for a lack of trying. Right.

It's hard to serve global audiences. and, you know, millions or hundreds of millions of people without sophisticated distillation attacks slipping through the cracks. And distill models are, can be dangerous because they likely lack safety guardrails, right, that could enable military in surveillance use, right? That's one of the arguments a lot of the AI labs are making, you know, as they start, you know, talking to global leaders, both global AI leaders and just political leaders as well, saying, hey, we have to be able to put a, you know, put clamps on this, right?

We have to be able to, you know, almost protect this as American trade secrets, as American IP, because if it gets into the wrong hands, it could, in theory, be used against the U.S. I do think that's obviously a very popular and prevailing argument about, at least when you start to look at the geopolitical aspects of AI and why AI companies here in the US want to be able to keep a little bit of a lid, right, obviously on their IP. And there's actually right now four congressional bills that are targeting chip exports, right? So that's another reason why we've seen a lot of these chip exports, you know, on Nvidia and

others, although they've gone back and forth and there's exceptions and loopholes and they're changing every week, right? But that's one of the reasons because U.S. lawmakers are like, we don't want to give, you know, the best technology to China or to certain other countries. But what we've seen here is they're finding a shortcut. They're finding a way around, which is actually probably faster. So, yeah, you know, maybe someone is sitting there is like, all right, well, you're going to, you know, prohibit the chips that we can buy. All right.

Well, we'll just distill the models. Much easier, much faster, much cheaper, right? than having to buy, you know, billions of dollars of GPUs and take the time to do it yourself. And it's not just Anthropic, right? And the frequency here is ramping up. So just in the last like two and a half weeks. So obviously Anthropic just published their findings yesterday. Oh, no.

So yeah, that was the 23rd. I'm wrong on my little date here on my slide. So on Monday, Open AI sent a memo to Congress naming deep seek on February 12th. So that was just under two weeks ago. And then similarly in February, Google identified over 100,000 distillation prompts targeting Gemini that same week. So just in the last two and a half weeks, the three big model makers have essentially gone public, right? But none as much as anthropic. Yes, Open AI did technically, you know, take, take their issues straight to Congress, but they didn't in the same way as Anthropic. Name names, right? They didn't come with

the receipts, right? And like I said, maybe they have more than they unveiled. But Anthropic really, really just kind of put it all out there on the table. Essentially, Open AI just alleged Deepseek use its models, outputs to train their rival R1 chatbot. Right? And Open AI just kind of said that they were free riding on American R&D. So I'm guessing that Open AI may have more specifics and there might be a reason that they didn't, you know, unveil all those. But Anthropic really came out with the receipts here. But it's happening to all of the big labs.

So like I said, Open AI said that several major Chinese LLM providers show distillation consistent patterns, but they only named deep seek by name. And they did describe networks of unauthorized. resellers that are helping Chinese labs evade access controls. So like I said, this is much, it's about much more than AI, right? This is about, you know, the governments between the two nations, other parties getting involved, you know, in helping, you know, allegedly helping Chinese labs skirt some of these restrictions. We've seen the same things with GPU chips as well. And so far, no response from the Chinese AI labs.

Deep Seek, according to reports, has not responded to a single media request across 14 months of allegations. Moonshot AI and mini-max declined every comment request. And then also the Chinese government and state media have offered no official response on the distillation claims when confronted by U.S. agencies. So what can be done? right because right now watermarking not really going to work great limiting not really going to work it's you know it's too easy to overcome those things at scale right in a model's usefulness and

its security just exists in a direct and unavoidable tradeoff right because you can access so much information via these APIs heck even just using something on the front end right with with these higher paid plans, you can get a group of researchers together, lock them in there, you know, or use an AI that does that. So it's not even like you need super sophisticated,

you know, systems to do this. You just need a little bit of money, some accounts. You know, if you're doing on the API side, it's much easier. But this is not necessarily something that requires a high level of

sophistication because literally it is the outputs and the thinking that you need. to train new models inputs. So here's why we need to pay attention. Do you remember back to January 2025? Do you remember what happened? Right? Deepseek model came out, just did really well on the benchmarks.

It sent economic shock through the markets, quite literally. Invidia alone lost more than Zinclair. $600 billion in a single day. Other tech stocks, I believe there was a total temporary loss of more than a trillion dollars, a trillion dollars because of something that ultimately was not true, right? Because all of a sudden, there is this panic, right? And I've been talking about this quite literally, man, the only good thing about having a daily podcast because it's kind of cringe when someone asks and I'm like, oh, I have a podcast.

That's my job is I have so many receipts. You can go see. I've been talking about this since late 20, 23. early 2024 about how essentially AI companies are driving not just the U.S. economy, but the world economy, right? For decades, you've never had such a concentrated mix of the top companies by market cap in the U.S. being all in the same industry or sector. It's actually quite literally never happened, right? But now you have the five biggest companies when it comes to market cap in the U.S. all essentially their AI companies are companies that have become

AI companies, right? You can make the claim like Microsoft, you know, Tesla, etc. Amazon, right? Google. But this hasn't happened before. So when you get a story like the deep seek, right? And they say, oh, yeah, we train this for, you know, pennies according to everyone else. You can see how that's going to send legit panic through everyone, right? If you're looking at your retirement and you're like, wow, you know, know, 30% of my retirement is in these seven companies, right? This isn't good. Oh my gosh, what's going to happen here?

Right. So cheap, open Chinese models may seem like a win. But like, what's the cost? Because distillation technically, I think, threatens the very business model of every U.S. company because these models powering American businesses. That's the other thing that we aren't really recognizing. Yes, some American businesses are opting to use Chinese models. whether they're doing so locally or doing so on the API side, which I personally would not recommend.

However, most enterprise companies in the U.S. have started to completely reshape how they run. And in the same way, the auto industry runs on fuel, runs on oil, right? the American enterprise runs on AI. I don't care what you say. Look at every single company that has continued to grow and that is pushing the U.S. economy forward. They are running on AI. They are implementing AI top to bottom.

So this is so much more than just, oh, a Silicon Valley issue. Because even the models right now that are powering these American businesses, well, they also may just be true. training their foreign competitors right now and simultaneously, yet slowly, eating away at public company gains. So what's the takeaway here? Well, the takeaway is even though this might just seem like some dorky AI drama, it's much more.

This is about the future of the who is the global superpower, right? This is about, right, who achieves artificial general intelligence first, right? Who is going to start blazing the path toward artificial superintelligence, which I don't even want to get into today. This is about more than copying the smart kids paper in front of you. This impacts every major U.S. business. This probably impacts your portfolio. This is something that deserves our attention. Do we know the answer? No. Is there a way that this could be solved? I'm actually not sure. But we're obviously going to be continuing to talk to smart people here on this show.

We're going to continue to bring you the information. And hopefully as answers become available on what's next in this saga, right? What's next in the U.S. proprietary models versus the Chinese open source models? We're not sure what's going to happen next, but we're going to be here every day trying to help provide you the answers and clarity on what to do next. All right. I hope this one was helpful. If it was, make sure you also go. out and check out that episode 712, 713, our 2026 AI prediction and roadmap series. And if this

was helpful, tell us something about it. Please take 30 seconds. Subscribe to the podcast. If you're listening on Spotify, Apple Podcasts, wherever you get your podcast, please take a couple of seconds to subscribe. That really means a lot to us and it helps us reach other people who are so confused about what the heck is going on in AI. All right. I'm going to try to help and give you the straight facts on that. So thank you for tuning in. Please go to your everyday AI.com. Sign up for the free daily newsletter. Thanks for listening today. We hope to hear you back tomorrow and every day for more everyday AI.

Thanks, y'all. Meet Firefly AI Assistant. Now live in Adobe Firefly, the Allman One Creative AI Studio. Just describe what you want to create in your own words and the assistant handles the rest. Orchestrating multi-step workflows across Adobe Creative Cloud apps, including Photoshop, Premier Express, and more in one conversational interface. You direct the outcome while the assistant accelerates execution. stand control with the ability to step in and refine at any time.

See it today at firefly.adobie.com. And that's a wrap for today's edition of Everyday AI. Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going. For a little more AI magic, visit Your EverydayAI.com and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next time.