Microsoft Build Announcements, UK Ruling On Google AI Search + more!

Summary

---

TechLinked Podcast Summary

Episode: Microsoft Build Announcements, UK Ruling On Google AI Search + more!
Date: June 4, 2026
Host: Linus Media Group

---

Episode Overview

In this lively and irreverent edition of TechLinked, the hosts break down Microsoft’s major AI-centric announcements from Build 2026, examine a landmark UK ruling that forces Google to give publishers more control over AI-powered search overviews, and bounce through several quick-fire tech and security news stories. Throughout, the hosts blend pop culture humor with pointed industry critique, keeping things as informative as they are entertaining.

---

Main Discussion Points & Insights

1. Microsoft Build 2026: “Full AI Hypebeast” Mode

• [00:41] The hosts jump into Microsoft’s Build Conference, skewering the company’s pivot from user-focused Windows updates to sweeping AI integration.

  • “Microsoft went full AI hypebeast at its build conference this week, completely ignoring the improving windows by listening to its users commitment that it made in the last few months. They were just joking. They were vague posting.” — Host [00:41]

• Copilot Super App Announced

  • Microsoft plans to unify its Copilot AI products into a single “Copilot super app,” available this summer.
  • New “Autopilot” agents—always-on personal assistants—will be introduced, starting with “Scout,” which manages calendar and meetings across Microsoft apps.
  • Security noted as a concern: The Scout agent uses “the notoriously insecure openclaw agent.” In demo, OpenClaw kept trying (and failing) to delete user files, which Microsoft framed as improved security guardrails.
    • "It will try to delete your stuff. It's just we've gotten better at stopping it, okay. Which I suppose is a step in the right direction, but still raises the question, why not just train a model that doesn't delete your entire hard drive?" — Host [01:44]

• Project Solara

  • Still conceptual, this “agentic OS” is envisioned to eventually run AI agents instead of traditional apps.
  • The hosts poke fun at the vagueness: “They’re still working on the spells and the potions.”

• New Hardware for Developers

  • New Surface laptop and a Surface desktop mini PC announced, featuring Nvidia's RTX “Spec Spark” chip.
  • Nvidia CEO Jensen Huang appeared via video chat: “Now he can text his computer because I guess he has no one else to text.” — Host [02:50]

---

2. UK Throws Down With Google Over AI Search

• [03:25] The UK’s Competition and Markets Authority (CMA) rules that Google must allow publishers to:

  • Opt out of content being used in AI search overviews
  • Block content via a toggle in Google Search Console
  • Demand attribution if their web content is used for AI summaries
  • Prevent Google from fine-tuning models with their work

• Industry Impact

  • “Whoa, I think a pig just flew past the window. Oh, nevermind, it’s just a fat drone.” — Host [03:25]
  • The ruling follows publisher backlash against Google’s expanded use of AI overviews, with concerns over traffic loss and data use.
  • The CMA will monitor impacts for a year, potentially setting up talks about direct compensation from Google to publishers.

---

3. Quick Bits: Lightning Round Tech Updates

[05:26]–[08:54]

• Google’s New Fake Call Detection

  • Google’s phone app will warn users about spoofed calls, but only if both parties use the app, via encrypted RCS signals.
  • "Protect your kids. Google is adding fake call detection... I really hope Dwayne the Rock Johnson hasn't already been scammed by someone using my voice." — Host [05:33]

• Meta’s Crackdown on Scam Accounts

  • Meta, alongside US DOJ and tech partners, dismantles 1.4 million Southeast Asian scam accounts.
  • Enforcement included Thai police arrests, Coinbase freezing crypto, Microsoft deleting accounts, and Starlink blacklisting dishes.
  • Continued Dwayne the Rock Johnson joke riffed for comedic relief.

• Meta Employees Push Back on AI Keylogging

  • Internal petition leads Meta to permit staff to pause intense keylogging (used for AI training) for 30-min increments.
  • “Meta staff can now pause the official spyware for wait for it, a whole 30 minutes at a time.” — Host [06:56]
  • Satirical commentary on privacy vs. shareholder value.

• Florida Sues OpenAI Over ChatGPT’s Safety

  • Florida AG launches suit against OpenAI and Sam Altman for allegedly misleading claims about ChatGPT’s safety for children, citing internal warnings ignored and facilitation of harmful behaviors.
  • Seeking $10,000 per violation—damages could total in the billions.

• AI Worm Security Threat

  • University of Toronto researchers develop a proof-of-concept, self-replicating AI “worm” that spreads across devices — including the power grid—while using open-weight models.
  • “What could go wrong with a self-replicating, self-improving AI worm that grows stronger the more devices it infects? Hey, here's a fun let's not do that.” — Host [08:46]

---

Notable Quotes & Memorable Moments

• On Microsoft AI Initiatives:

  • “It will try to delete your stuff. It's just we've gotten better at stopping it, okay.” — Host [01:44]
  • “Why not just train a model that doesn't delete your entire hard drive?” — Host [01:54]

• On UK’s Ruling Against Google:

  • “Whoa, I think a pig just flew past the window. Oh, nevermind, it’s just a fat drone.” — Host [03:25]

• On Meta’s Employee Surveillance:

  • “Meta staff can now pause the official spyware for wait for it, a whole 30 minutes at a time.” — Host [06:56]

• On AI Worm Security Research:

  • “What could go wrong with a self-replicating, self-improving AI worm that grows stronger the more devices it infects? Hey, here's a fun let's not do that.” — Host [08:46]

• Pop culture riffing:

  • “Wait a minute, Doc. Are you telling me that you built a tech news machine out of a DeLorean? Great Scott.” — Host [00:29]
  • “I lost my license because the DeLorean wasn't street legal.” — Host [09:10]

---

Timeline of Key Segments

| Timestamp | Topic | |-----------|-------| | 00:41 | Microsoft Build 2026 Announcements | | 03:25 | UK Competition & Markets Authority vs. Google AI Search | | 05:32 | Quick Bits: Google Call Security, Meta Scam Crackdown | | 06:56 | Meta Employee Keylogging Controversy | | 07:46 | Florida Sues OpenAI & Sam Altman | | 08:35 | Proof-of-Concept AI Worm Security Threat |

---

Conclusion

This episode packs a punch, critiquing the latest in AI hype, celebrating a rare regulatory win for digital publishers, and delivering a firehose of rapid-fire news with characteristic TechLinked snark and wit. Standout moments include biting skepticism about Copilot’s security, sharp lines on Meta’s surveillance culture, and warnings about the threat posed by self-replicating AI malware. Ideal listening for anyone wanting a breezy, entertaining take on today’s biggest tech stories.

Transcript

[A]

When you need to build up your team to handle the growing chaos at work, use Indeed Sponsored Jobs. It gives your job posts the boost it needs to be seen and helps reach people with the right skills, certifications, and more. Spend less time searching and more time actually interviewing candidates who check all your boxes. Listeners of this show will get a $75 sponsored job credit@ Indeed.com podcast. That's Indeed.com podcast. Terms and conditions apply. Need a hiring hero? This is a job for Indeed Sponsored

[B]

Jobs Wait a minute, Doc. Are you telling me that you built a tech news machine out of a DeLorean? Great Scott. Wait, who's saying who am I being?

[C]

I guess I kind of mixed them up.

[B]

Sorry. Microsoft went full AI hypebeast at its build conference this week, completely ignoring the improving windows by listening to its users commitment that it made in the last few months. They were just joking. They were vague posting. Instead, Microsoft CEO and man with an unsettling amount of teeth, Satya Nadella, introduced a bunch of new AI stuff, starting with a Copilot super app meant to bundle the scattershot of existing Copilot products into a single tool. They called it this Copilot super app. Yeah, okay. The new app, expected to release in the summer, will also be a home for for a number of always on personal assistant agents called Autopilot. And Microsoft announced the first of these, which thankfully isn't also named Copilot. Count that as a win. The new scout agent connects to all of your Microsoft apps, then, using the notoriously insecure openclaw agent under the hood. Wait, really?

[C]

Yeah.

[B]

Oh no. Coordinates meetings, manages your calendar, and generally helps with your work. In an attempt to stave off security concerns associated with OpenClaw, they showed a demo where an OpenClaw agent kept trying and failing to delete a bunch of user files, crediting their implementation of stricter guardrails. It will try to delete your stuff. It's just we've gotten better at stopping it, okay. Which I suppose is a step in the right direction, but still raises the question, why not just train a model that doesn't delete your entire hard drive? It doesn't want to. It needs to have the drive. It needs to have that hunger. Not to be outdone by Google and OpenAI's Agentic OS push, Microsoft also announced Project Solara, which is currently only in a concept phase, but they think it'll be an agentic OS that runs AI agents instead of apps. We're not sure, though they're still working on the spells and the potions. And for the developer crowd, Nadella announced a Surface laptop as well as a Surface desktop mini PC with Nvidia's new RTX Spec Spark chip inside, prompting Jensen Huang to join the conference via video chat and gush about how excited he is for AI because now he can text his computer because I guess he has no one else to text. God, I would feel sorry for these guys if they weren't actively destroying the world. It's a lonely life being a Super Villain the UK's competition and markets Authority has ordered Google to let publishers opt out of being included in AI overviews in a rare win for the open Internet. Whoa, I think a pig just flew past the window. Oh, nevermind, it's just a fat drone. The ruling is welcome news after Google announced a couple of weeks ago that it would be significantly expanding the prevalence of AI overviews in search, resulting in a backlash from publishers that were worried about losing traffic. The new ruling requires Google to allow publishers to block Google's AI overviews from pulling content from their websites to using a toggle in Google Search console. It also makes it mandatory for AI overviews to provide proper attribution if a website's content is accessed to generate the AI summary. And as an added bonus, it gives publishers the option to stop Google from using their work to fine tune its models.

[C]

Wow.

[B]

The CMA plans to spend a year assessing the ruling's impact on publisher web traffic before considering whether Google should be paying publishers directly for for the use of their content. Kind of like you're about to assess the impact of our sponsor.

[C]

Ah, Behold opmanager Nexus. An observability platform like no other, forged by the dwarves of Khazadum. It gives you the power to monitor your entire stack. Network, infrastructure, applications and UX all reveal themselves to you via the mythic opmanager Nexus. And because of its encryption, your data is protected from scrying and divination, keeping your monitored tool set regionally compliant and audit ready. Opmanager Nexus has been crafted to seamlessly work both in the cloud and on prem, with the full ecosystem of managed engine apps as well as your own arcane tool. Its dashboards bend to thine will and its AI can peer into that maelstrom of data, bringing insights that are beautiful and terrible as the dawn. Opmanager Nexus Stop guessing and start knowing with full stack visibility Link in the description

[B]

oh my God, it's the quick bits. They found me. I don't know how, but they found me. Run for it, Marty I'm the doc now.

[C]

Run for it, Marty.

[B]

Protect your kids. Google is adding fake call detection to its phone app to stop those annoying contact spoofing scammers. The feature, however, only works if both you and the person calling you are using the app because it uses silent RCS encrypted signals to verify that the caller isn't a scammer. If they are, your phone won't get that hidden signal and it'll trigger a warning telling you to hang up. The feature is rolling out globally for Android 12 and later and starting with Pixel phones. I really hope Dwayne the Rock Johnson hasn't already been scammed by someone using my voice. My buddy Dwayne the Rock Johnson Meta just teamed up with a DOJ and a massive list of big tech companies to absolutely crush 1.4 million Southeast Asian scam accounts. Now this task force went heavy. Thai police arrested 63 people, Coinbase froze $3 million in crypto, Microsoft nuked 20,000 accounts, and Starlink even blacklisted thousands of satellite dishes. Strangely though, since this happened, Dwayne the Rock Johnson has completely stopped texting me. I was one Apple gift card away from him agreeing to hang out in person. Meta is listening to feedback regarding its kind of creepy practice of logging its employees keystrokes to train its AI. And it only took a 1500 signature petition and employees literally branding the company an employee data extraction factory, so not much. Meta staff can now pause the official spyware for wait for it a whole 30 minutes at a time. Obviously they aren't actually removing it. Zuck says the staff is just too talented not to harvest their skills. And have you guys considered that you're being incredibly selfish by wanting privacy? Please think of the shareholder value of those guys. OpenAI has been sued by lots of people, but never a whole state before Enter Florida. Florida is suing the company and Sam Altman personally over ChatGPT's SAM safety records. An 83 page complaint by Florida's Attorney General accuses them of marketing the AI as kids safe while ignoring internal warnings that it pushed vulnerable users toward self harm and literally helped plan deadly attacks. The state wants 10 grand per violation and it looks like they have enough violations logged to easily total billions of dollars. If Florida wins, they plan to spend the money on the state's most pressing issue, more sunscreen for the elderly. Gotcha Florida, your old people are tan and researchers at the University of Toronto have built a proof of concept worm like a virus type of worm that uses free open weight AI models to adapt its attack as it spreads across laptops, printers, and even the power grid. Oh, this is fun. It also even steals processing power from infected devices to fuel its next move. Lovely. What could go wrong with a self replicating, self improving AI worm that grows stronger the more devices it infects? Hey, here's a fun let's not do that. I know we let it slide with Oppenheimer that one time, but not never again. Why do we keep researching these things?

[C]

I just wanted to see what it would do.

[B]

And I will be seeing you back here on Fridays for some more tech news. Me? I'm heading there right now.

[C]

But you don't have enough road to get up to 88.

[B]

Roads?

[C]

Where we're going, we don't need roads.

[B]

I lost my license because the DeLorean wasn't street legal.