Ivy: Every photo, email, and document you've trusted Google with for twenty years… is now officially on the table for ad targeting.
Marcus: I'm Marcus.
Ivy: And I'm Ivy.
Marcus: And this is Broke It, with your daily AI news for Sunday, October 25th, 2026.
Ivy: We've got three stories for you today, and… they’re big ones.
Marcus: Google is offering a perfect memory in exchange for perfectly targeted ads. We'll get into it.
Ivy: Then, the 'PHP-ocalypse'—how two tiny bugs nearly broke the entire internet.
Marcus: And finally, OpenAI's new Decisions API is here. It's super fast, but there is a big catch.
Marcus: Alright, let's start with Google. They just updated their privacy policy, and Ivy, you weren't exaggerating in the open, were you?
Ivy: Not at all. The new policy says they can use your personal data—from Gmail, Drive, Photos, all of it—to train their AI, and of course, target ads. They do give you a choice, though.
Marcus: Right. It only happens if you opt in to their new AI features. It's a trade-off. And honestly, the features sound incredible. An AI that knows your whole digital life, acting as a real personal assistant?
Ivy: Oh, I can imagine it, and it's terrifying. 'Personal assistant' is a nice way of saying 'ad machine that's read every private email I've ever sent.'
Marcus: But think of the upside! You're planning a trip, it sees your flight confirmation, your hotel booking, your old photos from that city—
Ivy: —and then hits you with ads for luggage, rental cars, and restaurants. Suddenly, there's no plausible deniability in your own digital house.
Marcus: You say it's the end, I say it's the beginning! It's the price for a perfect memory—a tool that finally, truly knows you.
Ivy: That's exactly what scares me. A tool that knows you, owned by a company that sells that knowledge. So for everyone listening: when Google asks you to 'enable new AI features,' just know you're also checking the box for 'new ad features.'
Marcus: So, choose wisely. Or, you know... choose the awesome features.
Ivy: Let's see how 'awesome' it is when your AI 'helpfully' mentions that secret purchase it found in your email to your spouse. Hard pass.
Ivy: Okay, next up, let's talk about the digital plumbing we all depend on. A new report is calling it the 'PHP-ocalypse.'
Marcus: Sounds dramatic, but they're not wrong. If you don't know, PHP is a programming language that runs something like 75% of the internet. We're talking WordPress, Facebook, Wikipedia... they all rely on it.
Ivy: And someone found two critical bugs that basically gave them a key to take over a huge chunk of those sites. Quietly. From anywhere in the world.
Marcus: It was like finding a skeleton key that opens three-quarters of the doors on Earth. The good news? The open-source community jumped on it and patched it fast. Crisis averted.
Ivy: This time. But the report's bigger point is that we're building these incredible, world-changing AI systems on top of digital foundations that are decades old and, frankly, starting to crack.
Marcus: But isn't that always how technology works? We build on what came before. It's iterative.
Ivy: Sure, but the stakes are so much higher now. It's one thing for your blog to go down. It's another thing if the AI managing a city's power grid is running on code with a known flaw.
Marcus: The takeaway here, especially for builders, is: patch your systems. And maybe think hard about the foundation you're building that next billion-dollar AI company on.
Ivy: Bottom line: the digital world is way more fragile than it looks. This wasn't the first big scare, and it definitely won't be the last.
Marcus: Alright, our last story is about something new from OpenAI. They've rolled out their Decisions API to more people.
Ivy: Ah, yes. The promise of fast, easy answers. You feed it data, it spits out a label. Simple.
Marcus: And it is fast. I've been playing with it for stuff like content moderation or sentiment analysis... it's nearly instant. Apparently Uber is already using it for routing help.
Ivy: And how does it do that?
Marcus: Well... you send the data, and it sends back a decision.
Ivy: Exactly. It's a total black box. You have no idea what's happening inside, what biases are baked in, or how you'd ever debug it when it goes wrong.
Marcus: For a full production system, sure, that matters. But just for getting an idea off the ground? It's a huge shortcut.
Ivy: And then there's the other shoe: the cost. The pricing is apparently hard to predict. It's not a simple per-use fee. How can you build a business on something when you can't even budget for it?
Marcus: Okay, that's fair. The path to scaling up isn't clear at all.
Ivy: So, the takeaway is, while this Decisions API is a powerful toy for a prototype, you should think very carefully before building a real product on it. If you can't explain it and you can't budget for it, you probably shouldn't build on it.
Marcus: Harsh. I still think it's a game-changer for quick experiments.
Ivy: Experiments don't have to answer to shareholders. Let's see how they feel when the first bill comes due.
Marcus: Alright, let's recap the day.
Ivy: Google is offering you powerful AI tools in exchange for your entire digital soul.
Marcus: The internet is a fragile tower of old code that almost came crashing down this week.
Ivy: And OpenAI will make your business decisions for you—as long as you don't ask how, or how much.
Marcus: Hey, before we go... speaking of black boxes, I saw an AI art generator that just makes 'liminal spaces.' You know, empty hallways, abandoned malls...
Ivy: So it makes pictures of the internet before all the people arrived? ...Actually, that sounds kind of nice.
Marcus: Right? A little digital calm after a day of chaos.
Marcus: And that's our show. Thanks for listening to Broke It.
Ivy: We'll be back tomorrow. In the meantime, maybe go take a look at your sent mail. You know, just in case.
This show is made with AI: the hosts’ voices are synthetic and the scripts are AI-assisted. Every story links to its original source.