azcomputerguru/claudetools
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
604c9d9d4bd2a59456d2d7ffe7f6d55b148d69f1
claudetools/projects/radio-show/episodes/2026-03-21-who-controls-your-tech/show-prep.md
Mike Swanson 604c9d9d4b Session log: repo reorganization, GrepAI test, radio show prep 
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-20 18:34:11 -07:00

19 KiB
Raw Blame History

The Computer Guru Show - Show Prep

"Who's Really In Control?" - March 21, 2026

~90 minutes of content (6 segments + intro/outro across 2 hours with breaks)

Host: Mike Swanson Theme: Every major tech story this week comes back to one question: Who's really in control of your technology -- you, the companies, or the government?

Segment 1: "The Week That Was" (~12 min)

Theme: Set the table -- this was a massive week in tech

Opening

This was one of those rare weeks where every major story in tech connects to the same question. The White House dropped a national AI policy framework. NVIDIA held its annual GTC conference and basically declared itself the center of the AI universe. Apple quietly confirmed Google is going to power Siri's brain. A Canadian company lost a PETABYTE of data to hackers. And right to repair laws just kicked in for a quarter of Americans.

What do all of these have in common? They're all about control. Who controls the AI? Who controls the chips that run the AI? Who controls your phone? Who controls your data? And who controls whether you can fix your own stuff?

That's what today's show is about. Buckle up.

Quick Headlines Rundown

  • White House AI framework dropped TODAY (March 20) -- 7 pillars, wants to preempt state laws
  • NVIDIA GTC this week -- Vera Rubin chips, $1 trillion in orders, Disney robots on stage
  • Apple + Google deal -- Siri will run on Google Gemini
  • TELUS Digital breach -- 1 petabyte stolen by ShinyHunters
  • GPT-5.4 launched March 5 -- 1M token context, computer use built in
  • Right to Repair -- now law in 6+ states, covering 26% of Americans

Transition

"Let's start at the top. The President of the United States just told us how he wants to regulate AI. And whether you love it or hate it, if you use any technology at all, this affects you."

Segment 2: "The Government Wants In" (~15 min)

Theme: White House AI Framework -- what it actually says and what it means for regular people

Key Points

What happened: The White House released a "National Policy Framework for Artificial Intelligence" on March 20 -- literally today. This is the administration's blueprint for how Congress should regulate AI.

The 7 Pillars (plain English):

  1. Protecting kids -- No collecting data on children, parental controls required
  2. Protecting communities -- Don't let AI scam people, address deepfakes
  3. Copyright and creators -- They're saying AI scraping the internet is NOT copyright violation (this is huge and controversial)
  4. Free speech -- AI can't be used for censorship (aimed at platforms)
  5. American dominance -- We need to win the AI race, period
  6. Workforce -- Retrain workers, tax breaks for small businesses adopting AI
  7. Federal preemption -- This is the big one: override state AI laws with a single national standard

The real story -- preemption:

  • Washington state just passed 5 AI bills (chatbot safety for kids, AI in health insurance decisions, deepfake rules, digital likeness protection)
  • Colorado has the broadest digital repair rights in the country as of January
  • The White House framework wants Congress to sweep all of that away with a "light touch" federal standard
  • This is the same playbook we've seen with data privacy -- states act, then industry pushes for weaker federal law to override them

Why this matters to listeners:

  • If your state passed strong AI protections, they could be overridden
  • The copyright stance means your creative work can be scraped by AI without compensation (unless Congress creates a licensing mechanism)
  • Data centers may be allowed to generate their own power on-site -- good for AI speed, but what about your utility rates?
  • "Light touch regulation" historically means the companies regulate themselves

Listener Q&A Prep

"Is this good or bad?" It's complicated. Protecting kids and fighting scams -- good. Preempting stronger state laws -- debatable. Saying AI can scrape copyrighted content freely -- creators are furious. The "light touch" approach works until it doesn't.

"Will this actually become law?" Probably not before midterms in November. But it signals where the administration wants to go. And it gives AI companies a talking point: "Don't regulate us at the state level, the feds are handling it."

Transition

"So the government is trying to figure out how to regulate AI. Meanwhile, one company is trying to make sure that no matter what AI you use, it runs on THEIR chips..."

Segment 3: "Jensen Huang's Trillion-Dollar Bet" (~15 min)

Theme: NVIDIA GTC and the question of who controls the AI supply chain

Key Points

What happened at GTC (March 16-19): Jensen Huang delivered his keynote in San Jose and basically laid out NVIDIA's vision for owning every layer of the AI stack.

The big announcements:

  • Vera Rubin NVLink 72 -- next-gen GPU system, the "engine supercharging the era of AI"
  • $1 trillion in orders -- Huang said he expects purchase orders between Blackwell and Vera Rubin to hit $1 trillion through 2027. Let that number sink in.
  • Groq 3 LPU -- NVIDIA unveiled a chip from Groq, the startup they acquired for $20 billion in December. This is an inference chip -- it runs AI models, not trains them.
  • NemoClaw -- a reference stack for building AI agents. Huang's pitch: "It builds you an AI agent."
  • Uber autonomous fleet -- 28 cities, 4 continents by 2028, powered by NVIDIA Drive AV
  • Disney's Olaf walked on stage -- an autonomous robot showcasing physical AI

The control angle:

  • NVIDIA doesn't just make GPUs anymore. They're building the chips, the software stack, the agent frameworks, AND the autonomous vehicle platform.
  • Every major AI model -- GPT-5.4, Claude, Gemini -- trains on NVIDIA hardware
  • Eli Lilly just built a supercomputer with 1,016 NVIDIA Blackwell GPUs to design drugs
  • When one company controls the picks and shovels of an entire technological revolution, that's enormous power
  • Jensen Huang is arguably the most powerful person in tech right now -- not because he makes the AI, but because nothing works without his chips

The Groq acquisition angle:

  • NVIDIA bought Groq for $20 billion and just unveiled its chip at GTC
  • This means NVIDIA now controls both training AND inference chips
  • Training is how AI learns. Inference is how AI thinks when you use it. NVIDIA just locked up both sides.

For regular people:

  • The cost of AI is going up, not down -- and NVIDIA's pricing power is a big reason
  • When your AI assistant responds slowly or your company can't afford better AI tools, part of that traces back to GPU supply and pricing
  • This is like if one company made every engine for every car, truck, and bus on the road

Transition

"So NVIDIA controls the hardware. But here's where it gets really interesting for iPhone users. Apple just basically admitted they can't build their own AI brain. So who did they call? Google."

Segment 4: "Apple Gives Google the Keys to Siri" (~15 min)

Theme: The Apple-Google AI deal and what it means when you don't control your own product's intelligence

Key Points

What happened: Apple and Google announced a multi-year collaboration. The next generation of Apple's AI will be based on Google's Gemini models -- specifically their 1.2 trillion parameter model. A "more personalized Siri" is coming with iOS 26.4, expected this spring.

Why this is a big deal:

  • Apple has spent YEARS talking about doing everything in-house -- "designed in California" is their whole brand
  • They tried to build their own AI. It wasn't good enough.
  • Now the company that built its reputation on privacy is handing your Siri conversations to Google's AI infrastructure
  • Yes, it'll run on Apple's "Private Cloud Compute" -- but the MODEL is Google's

The privacy question:

  • Apple's pitch has always been: "What happens on your iPhone stays on your iPhone"
  • Google's entire business model is built on knowing everything about you
  • Apple says privacy is maintained through their cloud compute infrastructure
  • But the underlying intelligence -- the thing that understands your question and generates an answer -- is Google technology
  • How comfortable should you be with that?

The business angle:

  • While Microsoft, Google, Meta, and Amazon are pouring tens of billions into building AI infrastructure, Apple is taking a different approach
  • Apple is turning AI into revenue through features, subscriptions, and device upgrades
  • They're letting Google spend the billions on training models, then licensing the result
  • Smart business? Maybe. But it means Apple users are now dependent on Google's AI in a way they never were before.

OpenAI loses:

  • OpenAI was reportedly also bidding for this deal
  • Google winning means Gemini gets access to Apple's massive install base -- over a billion active devices
  • This reshapes the entire AI competitive landscape

For listeners:

  • If you have an iPhone, your Siri is about to get a LOT smarter
  • But the brain behind it will be Google's, not Apple's
  • Ask yourself: when you chose Apple for privacy, did you sign up for Google processing your AI requests?

Transition

"So we've got the government trying to regulate AI, NVIDIA controlling the chips, and Google powering Apple's AI. But all of that assumes your data is safe in the first place. And this next story is a reminder that it very much is not."

Segment 5: "A Petabyte of Your Data, Gone" (~15 min)

Theme: The TELUS Digital breach and the state of cybersecurity in 2026

Key Points

What happened: TELUS Digital -- a major outsourcing company that handles customer service for big brands -- confirmed on March 12 that hackers stole nearly 1 PETABYTE of data. That's roughly 1,000 terabytes. For context, the entire Library of Congress digitized collection is about 20 petabytes. This is an absolutely staggering amount of data.

Who did it: ShinyHunters, a well-known hacking group. They demanded $65 million ransom. TELUS is reportedly refusing to pay or even communicate with them.

What was stolen:

  • Customer data from TELUS Digital's BPO (business process outsourcing) clients
  • Source code
  • FBI background check data (yes, really)
  • Financial information
  • Voice recordings of customer service calls
  • Salesforce data for multiple companies

How they got in (this is the scary part):

  • ShinyHunters found Google Cloud Platform credentials that were leaked in a DIFFERENT breach -- the Salesloft/Drift breach
  • They used a tool called TruffleHog to search those leaked credentials for more passwords
  • Those passwords let them pivot into TELUS systems
  • One breach led to another breach. Your data's security is only as strong as the weakest company in the chain.

The "control" angle:

  • When you call customer service for Company X, and they outsource to TELUS Digital, you didn't choose to give TELUS your data
  • But TELUS had your voice recordings, your financial info, potentially your background check
  • You had zero control over that decision and zero visibility into their security
  • This is the hidden supply chain of your personal data

Other breaches this month:

  • Navia Benefit Solutions -- 2.7 million people's health benefit data exposed
  • Kaplan North America -- 192,000 individuals' personal info compromised
  • Iran-linked hackers used Microsoft Intune to remotely wipe thousands of employee phones at medical device maker Stryker

Critical vulnerability alert:

  • SharePoint CVE-2026-20963 -- patched in January, still being actively exploited
  • If your workplace uses SharePoint and hasn't patched: sound the alarm
  • PolyShell vulnerability affects ALL Magento/Adobe Commerce installations -- unauthenticated code execution

Practical advice for listeners:

  1. You can't control who companies share your data with, but you CAN limit what you give them
  2. Use unique passwords everywhere -- credential stuffing from one breach causes the next
  3. If a company you do business with has a breach, don't wait for their notification. Change passwords immediately.
  4. Freeze your credit if you haven't already. It's free and reversible.
  5. Voice recordings being stolen means voice-clone scams are going to get more convincing

Transition

"So companies can't protect your data, the government is just now figuring out how to regulate AI, and the tech giants are consolidating control over your devices. Is there any good news? Actually, yes. For the first time, a quarter of Americans now have the legal right to fix their own stuff."

Segment 6: "Taking Back Control" (~15 min)

Theme: Right to Repair, digital ownership, and subscription fatigue -- the fight to actually own your technology

Key Points

Right to Repair milestone: As of January 1, 2026, more than 25% of Americans live in a state with Right to Repair protections. By fall (Connecticut in July, Texas in September), that rises to over 35%.

Colorado leads the way:

  • Broadest digital repair rights in the nation as of January 2026
  • Manufacturers MUST provide independent repair shops AND consumers with tools, parts, and documentation
  • Key new protection: manufacturers cannot use software locks to prevent component replacement
  • This directly targets Apple's "parts pairing" -- where replacing a screen with a genuine part still triggers warnings or disables features because the software doesn't recognize it

The software lock problem:

  • You buy a tractor from John Deere -- but John Deere's software controls whether it runs
  • You buy a phone from Apple -- but replacing a cracked screen might disable Face ID
  • You buy a smart refrigerator -- but the manufacturer can brick it through a software update
  • "Ownership" increasingly means "you paid for it but we still control it"

Hardware-as-a-Service (HaaS):

  • Growing trend: companies want you to subscribe to hardware, not buy it
  • Your printer, your security system, your smart home devices -- all becoming subscription services
  • The risk: when the subscription ends, physically working devices become unusable
  • We're heading toward a world where you don't own anything -- you just rent access

Subscription fatigue is real:

  • Average American household now has 7+ subscription services
  • Add hardware subscriptions on top of streaming, software, cloud storage, and apps
  • People are hitting a wall -- "subscription fatigue" is driving a counter-movement toward ownership
  • There's a growing market for "buy once, own forever" products

The thread that ties it all together:

  • The White House wants to preempt state repair and digital rights laws
  • Companies like Apple are moving MORE of your phone's intelligence to the cloud (Google's cloud)
  • Data breaches prove that the more companies hold, the more they lose
  • Right to Repair says: if I paid for it, I should be able to fix it, modify it, and control it

What listeners can do:

  1. Support Right to Repair legislation in your state (Arizona doesn't have a law yet!)
  2. When buying tech, ask: "Can I fix this myself? Will it work without an internet connection? What happens if the company goes under?"
  3. Consider devices and software that respect ownership: Linux, Framework laptops, Fairphone, local-first software
  4. Push back on "subscription-only" products. Vote with your wallet.
  5. Back up your data locally -- don't depend entirely on cloud services you don't control

Transition to close

"And that really is the theme of everything we talked about today..."

Closing Segment (~3 min)

Theme: Bring it all home

Wrap-up

We covered a lot of ground today, but it all came back to one question: who's in control?

The government is trying to take control of AI regulation -- but in a way that might actually give companies MORE freedom, not less. NVIDIA controls the chips that make AI possible, and they just tightened their grip. Apple handed control of Siri's brain to Google. Hackers proved -- again -- that the companies we trust with our data can't always control who else gets it.

But here's the thing -- and this is what I want you to take away today: you still have choices. You can choose what data you share. You can support laws that protect your right to repair and own your devices. You can ask hard questions before buying into the next subscription service.

Technology is incredible. I wouldn't be sitting behind this microphone if I didn't believe that. But incredible technology in someone else's control is just a prettier cage. The goal should always be: technology that empowers YOU.

That's the show. I'm Mike Swanson, the Computer Guru. Thanks for listening. We'll see you next time.

Bonus: Caller/Segment Fillers

If segments run short or you get calls, here are quick-hit stories to fill:

  1. Eli Lilly's AI Supercomputer -- 1,016 NVIDIA GPUs, trying to cut drug development from 10 years to 5. Is AI-designed medicine something you'd trust?

  2. Hyundai + Boston Dynamics robots -- "AI+Robotics" roadmap for logistics and personal assistance. Your next delivery might come from a robot dog.

  3. Tennis-playing humanoid robot -- Galbot's Unitree G1 trained on 5 hours of data, 96% forehand success. When does AI start competing in sports?

  4. Disney's robot Olaf at GTC -- Walked on stage, fully autonomous. The future of theme parks is AI characters that improvise.

  5. GNU Telnet vulnerability (CVE-2026-32746) -- 9.8/10 severity score. Ancient protocols still haunt us. "If you're still using telnet, please stop."

  6. Claude gets memory -- Anthropic rolled out persistent memory for Claude in early March. Your AI assistant now remembers your preferences across conversations. Convenient or creepy?

Research Sources

Reference in New Issue View Git Blame Copy Permalink