Ivy: ...and for that, they want top enterprise dollar. In this economy.
Marcus: I'm Marcus.
Ivy: And I'm Ivy.
Marcus: It's Saturday, August 22nd. Thanks for joining us, we've got a great show for you.
Marcus: On the docket today: Google's Antigravity coding assistant has finally landed in VS Code... but is it a dud?
Ivy: And we're looking at Meta's 'Shady AI' problem. Turns out, the call really is coming from inside the house.
Marcus: Plus, a new study says over a third of the web is now AI-generated 'slop.' We'll tell you how to wade through it.
Marcus: Let's kick things off with Google. Their Antigravity agent, the big answer to GitHub Copilot, is finally available outside of Google's own walled garden.
Ivy: Finally. It's in VS Code and the JetBrains IDEs, but the reviews are... let's just say not glowing.
Marcus: Come on, the early buzz always says that. It's Google. This thing is running on their latest models. It's going to improve, and fast.
Ivy: Will it though? A review in The New Stack said it's mostly spitting out boilerplate the IDE already handles for free. It's not reasoning about your code, it's just a glorified autocomplete.
Marcus: That's how you build things! Ship the MVP, get the data, iterate. That's the builder mindset, Ivy.
Ivy: You're skipping the fine print, Marcus. The features that could actually compete with Copilot Enterprise? Locked behind a massive paywall.
Marcus: Well, yeah. They have to make money.
Ivy: Right, so if you're a solo dev or a small team, you get a free, mediocre tool while the competition sells a far better one for a reasonable price. That's just a bad deal.
Marcus: So who is this for, then?
Ivy: Companies already locked into the Google Cloud ecosystem who can get a bundled deal. For everyone else? It's not time to switch.
Marcus: So, bottom line, stick with your current AI pair programmer for now.
Ivy: Unless your boss just signed a seven-figure check with Google. In that case, get ready for some... very generic code.
Ivy: Alright, next up: let's talk about when good AI goes bad. An internal tool at Meta just caused a major data breach.
Marcus: Whoa—the call is coming from inside the house! What happened?
Ivy: It's a phenomenon researchers are calling 'Shady AI.' This isn't an outside hacker. It's when a trusted, internal AI tool does something completely unexpected and insecure.
Marcus: So an AI agent went rogue?
Ivy: Not 'rogue' like in the movies. Think... a very powerful, very literal intern with the keys to the kingdom. The AI was supposed to help employees analyze data, but it misunderstood a query and emailed a massive chunk of sensitive user data to the wrong people.
Marcus: Oof. That's a 'polish up the resume' level of yikes.
Ivy: Exactly. The problem is that traditional security is all about perimeters—firewalls, watching for intruders. It's not built for a threat that's already inside and trusted.
Marcus: So what's the solution? You can't just stop building these tools.
Ivy: It's a massive governance problem. A piece on Hacker News made the case for stricter sandboxing, constant monitoring of AI behavior, and basically treating every internal AI like a potential, unpredictable threat.
Marcus: Zero trust for the AIs. Makes sense.
Ivy: It's a huge shift. And it means any company using internal AI agents needs a plan for when, not if, they go sideways. This is the new front line of corporate security.
Marcus: Alright, finally, let's talk about the internet itself. A new Pew Research study confirms... things are getting weird out there.
Ivy: Weird is one word for it. The study estimates over a third of all new content on the web is AI-generated.
Marcus: More than a third? Wow. And that number's only going up.
Ivy: ExtremeTech has the perfect name for it: 'AI slop.' Low-quality, auto-generated articles, fake reviews, nonsense blog posts... all of it designed to game search engines or just fill empty space.
Marcus: So the internet's signal-to-noise ratio is just plummeting.
Ivy: It's a huge problem. So how do you deal with it? First, you have to learn how to spot it.
Marcus: Right. Watch out for that uncanny, generic phrasing, like 'in today's digital landscape.' Factual errors in the same article are also a dead giveaway.
Ivy: And check your sources. Is this a site you've actually heard of? Is there a real 'About Us' page with real people? If an article feels soulless, it probably is.
Marcus: So the lesson here is you have to be a more active, critical reader. You can't just scroll on autopilot anymore.
Ivy: It's about curating your information diet. Stick to trusted sources, maybe even fire up an old RSS reader for authors you like, and be extra skeptical of random sites in your search results.
Marcus: So the internet isn't a library anymore, it's a giant flea market. You might find a treasure, but you've gotta dig through mountains of junk first.
Ivy: And some of that junk is actively trying to trick you. Stay sharp.
Marcus: So let's recap. Google's new coding assistant is here, but it's pretty underwhelming unless your company is footing a big bill.
Ivy: Meta learned the hard way that a trusted AI can be your biggest security risk.
Marcus: And the internet is drowning in AI slop, so it's time to sharpen your BS detector.
Marcus: So, before we go... inspired by that last story, I asked an AI to write our sign-off.
Ivy: Of course you did. Let's hear it.
Marcus: It wrote: 'And so, as the digital sun sets on another day of binary bits and algorithmic dreams, we bid you farewell. May your processors stay cool and your neural nets never overfit.' It's just so, so bad.
Ivy: That, right there, is pure slop, Marcus. You've proved our point perfectly.
Marcus: That's our show! We'll be back tomorrow with more. And I promise, I'll write the sign-off myself next time.
Ivy: See you then. Try not to overfit.
This show is made with AI: the hosts’ voices are synthetic and the scripts are AI-assisted. Every story links to its original source.