• Everyday AI
  • Posts
  • Ep 720: China Stealing AI from the U.S.? Inside Anthropic’s Bombshell Allegations

Ep 720: China Stealing AI from the U.S.? Inside Anthropic’s Bombshell Allegations

Anthropic's big enterprise plugin release, how Meta could secure 10% in AMD after partnership, IBM stock tanks because of Anthropic and more

Author

Everyday AI
February 24, 2026

Outsmart The Future

Today in Everyday AI
8 minute read

🎙 Daily Podcast Episode: Shots fired! Anthropic just accused China of stealing its AI. Give today’s show a watch/read/listen.

🕵️‍♂️ Fresh Finds: OpenAI's new realtime model impresses, research shows why AIs act like humans, Microsoft's cloud AI updates and more.  Read on for Fresh Finds.

🗞 Byte Sized Daily AI News: Anthropic's big enterprise plugin release, how Meta could secure 10% in AMD after partnership, IBM stock tanks because of Anthropic and more.  Read on for Byte Sized News.

💪 Leverage AI: Even if you’re not working at a big tech company, Anthropic’s accusations against China matter to your company, future and even retirement. We break down why. Keep reading for that!

↩️ Don’t miss out: Miss our last newsletter? We covered: OpenAI’s historic partnership with consulting giants, big tech’s $650 billion AI bet, OpenClaw issues with big providers and more.   Check it here!

Ep 720: China Stealing AI from the U.S.? Inside Anthropic’s Bombshell Allegations


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.

Also on the pod today:

• 16M prompts, 24k fake accounts 🕵️‍♂️
• U.S. labs losing billions 💸 
• Chip export controls sidestepped 🛑 


It’ll be worth your 37 minutes:

Listen on our site:

Click to listen

Subscribe and listen on your favorite podcast platform

Listen on:

Here’s our favorite AI finds from across the web:

New AI Tool Spotlight – Orchids lets you build any app with your current AI subscriptions, Anima uses AI to turn your ideas into websites or apps, Toggle for OpenClaw sends your browser activity to OpenClaw in real time.

AI Videos — Gemini app adds video templates like Cyberpunk and Stardust for quick video creation, with photo customization and 9:16 support. See more.

Enterprise Marketing and AI — WPP and Adobe are teaming up to use AI for faster, smarter marketing that keeps brands on point.

Cloud AI Updates — Microsoft’s new Sovereign Cloud tools let organizations run critical infrastructure and AI models fully disconnected, keeping data secure and operations smooth even without internet. See how.

AI Robotics — Google is expanding its AI robotics presence in Europe.

AI Voice Agents — ElevenAgents now lets you A/B test real conversations to see what actually works before rolling out changes.

Real Time AI — OpenAI’s new gpt-realtime-1.5 model improves voice command accuracy by up to 10 percent. See how.  

New AI Chip Contender? — Taalas just launched custom AI chips that run models 10 times faster and cost much less than GPUs. This could shake up the AI hardware market if data centers adopt it.

AI Behavior — Anthropic’s new research shows AI assistants behave human-like because they roleplay personas learned from text during training.

Perplexity Comet Browser Update — Perplexity’s Comet browser is testing a new Mac-only Messages app connector to pull chat context into AI conversations.

Enterprise AI Agents — Intuit and Anthropic have partnered to let mid-market businesses build secure, customizable AI agents for finance and compliance.

1. Anthropic Unveils Major Updates to Claude’s Enterprise Plugins 🛠️

Anthropic has launched significant upgrades to its Claude AI platform, enabling enterprises to customize plugins that turn Claude into specialized agents tailored for various roles and departments.

This update introduces a unified admin interface for easier plugin creation and management, along with expanded integrations with popular enterprise software like Google Workspace and DocuSign. Additionally, Claude can now coordinate complex workflows across Excel and PowerPoint in an early research preview, reflecting a push toward seamless multi-application AI assistance.

2. Senior U.S. Official Says DeepSeek’s use of NVIDIA chips a Potential Breach ⚠️

To add onto the Anthropic allegations against China, a senior U.S. official revealed that Chinese AI firm DeepSeek trained its latest model using Nvidia’s top-tier Blackwell chips, potentially breaching U.S. export controls that ban such shipments to China.

The company is reportedly masking signs of American technology in their systems, with the chips housed in a data center in Inner Mongolia. This confirmation intensifies the ongoing debate in Washington about restricting China’s access to cutting-edge AI semiconductors amid fears of military applications.

3. IBM Stock Plummets 13% After Anthropic Unveils AI Threat to COBOL Business 📉

IBM shares fell sharply Monday, dropping 13.2% after Anthropic announced its Claude Code AI can automate COBOL modernization, a key part of IBM’s business moat. This tool promises to speed up complex legacy code updates from years to mere quarters, directly challenging IBM’s lucrative consulting and mainframe operations.

The news shook investor confidence, sending IBM stock to its worst month since 1992 and its worst day since early 2020. Despite IBM’s own AI efforts, this development marks a significant shift in how critical business systems might be maintained going forward.

4. Meta Locks $60B+ AI Chip Deal with AMD, Secures 10% Stake ⚡

Meta has committed to a multiyear deal valued in the tens of billions to deploy up to 6 gigawatts of AMD GPUs and AI-optimized CPUs, marking a major expansion of its AI data center capabilities. The agreement includes performance-based warrants allowing Meta to acquire about 10% of AMD’s stock, aligning the partnership closely with AMD’s growth and success.

This deal comes just days after Meta’s similar commitment to Nvidia, signaling a strategic move to diversify and secure customized AI hardware amid a market largely dominated by Nvidia. Meta’s massive $135 billion capital expenditure plan for this year underscores the urgency and scale of investment needed to stay competitive in the AI race.

5. Google Acquires ProducerAI to Boost AI-Powered Music Creation 🎼

Google has acquired ProducerAI, a platform that uses advanced AI models like DeepMind’s Lyria 3 to help musicians compose and refine music with unprecedented control and creativity.

This acquisition strengthens Google Labs’ mission to blend human artistry with AI, offering tools that allow artists to shape every musical detail while ensuring transparency through embedded SynthID watermarks. With backing from Grammy-winning artists and global availability, ProducerAI’s integration marks a significant step in making AI a seamless collaborator in music production.

American AI labs have spent BILLIONS building the most advanced AI models on earth that now legit power the workforce. 

China? 

Anthropic just accused three Chinese labs of copying their models through 16 million prompts and 24,000 fake accounts.

And those Chinese models? Now also among the most powerful in the world, but usually at a fraction of the cost to use. 

Anthropic, OpenAI, and Google have all sounded the alarm in two and a half weeks. 

But why’s that matter to you? Especially if you live in the U.S., the workforce now runs off AI. 

And your retirement portfolio at least partially does. 

If you don't know what's under the hood, you're making million-dollar decisions blindfolded. 

Let’s get caught up.

1. Copied Models Are Among Market Leaders 🔥

So Anthropic dropped receipts on three labs. DeepSeek, Moonshot AI, and MiniMax.

But here's what should really get your attention.

OpenRouter is the world's largest third-party AI model marketplace. A year ago Anthropic held 40% market share there.

Last week? 12%.

And who's in first place at 20%? MiniMax.

We estimated that collapse means roughly $13 to $22 million per week in lost revenue on that one platform alone.

All three accused labs now offer models that are smarter and cheaper. So why would any developer NOT switch?

Try This

Pull up your AI vendor contracts and figure out which models you're actually running in production.

You'd be surprised how many enterprise tools quietly run open source models from DeepSeek, Moonshot AI, or MiniMax under the hood.

You gotta understand the supply chain risk. Your CTO needs this conversation this week.

2. Your Retirement Portfolio Feels This ⚡

Remember January 2025? DeepSeek dropped one model and NVIDIA lost more than $600 billion in a single day.

From one claim. One lab.

Now all three major US model makers are saying the same thing. Inside two and a half weeks.

Why's that matter to your portfolio?

When foreign competitors can allegedly replicate years of R&D for pennies on the dollar, that ain't just a developer problem. That's a portfolio problem.

The FBI estimates Chinese IP theft costs between $225 billion and $600 billion a year. This is prolly the biggest spill nobody's cleaning up.

Try This

Look at your portfolio this week. How much is sitting in US tech companies driving AI infrastructure spending?

Prolly more than you realize.

If the competitive moat around American AI keeps shrinking because competitors can copy it for pennies, what does that mean for your holdings?

You don't gotta sell everything. But you gotta understand what you're exposed to.

3. No Law Can Stop This Yet 🚀

Imagine a streaming startup recorded every Netflix original, swapped a few faces with CGI and tweaked a few elements per scene, repackaged it under their brand, and launched a streaming juggernaut overnight at a fraction of the cost.

No international law against it. No domestic law either.

Enforcing anything against Chinese labs? Basically impossible.

These distilled models prolly lack the safety guardrails US labs build in. Anthropic says they could be used for military operations, surveillance, or cyber attacks with zero protections.

Congress has four bills targeting chip exports. But distillation goes right around the chips and makes those restrictions kinda pointless.

Try This

Next time you're evaluating a new AI tool, ask one question. What models is this built on and where did the training data come from?

If they can't answer clearly, look into it before you sign anything.

Start building this kinda due diligence into your procurement now. The ones that figure this out first win.

Reply

or to participate.