Priya: ...and now Meta promises it wasn't going to scrape everything you post publicly to train its AI. Promise.
Theo: I'm Theo.
Priya: And I'm Priya. It's Sunday, October 18th, 2026, and you're listening to Money Daily.
Theo: Today, we're tracking the money behind the machines.
Theo: Meta is doing an about-face on some very scary-sounding terms of service.
Priya: A popular coding platform just picked a side in the AI model wars… and it's a big one.
Theo: And a former Nvidia insider explains why your robot butler still hasn't shown up.
Theo: So, if you use Facebook or Instagram, you've probably been worried about your public posts becoming robot food.
Priya: 'Worried' is putting it lightly. Meta's terms of service basically gave them the right to scrape all your public info to train their AI models.
Theo: Right, the classic 'your data is our data' fine print. But they've changed it! After some public outcry, TechCrunch reports Meta removed that specific language.
Priya: Hold on. They removed the language. A Meta spokesperson insisted, and I quote, that the update did not reflect a change in the way Meta crawls the web.
Theo: So... they're just not saying the quiet part out loud anymore?
Priya: That's my take. It's a masterclass in corporate PR. They aren't promising to stop doing it, they're just promising to stop telling you they're doing it in the terms you never read.
Theo: But isn't this still a win for users? The scary clause is gone. It shows they're responsive to pressure.
Priya: This is a win for their plausible deniability. The spokesperson's statement is key. They're telling us nothing has functionally changed. They're just cleaning up the evidence.
Theo: Wow. So what does this mean for the picture of my lunch I posted on Threads?
Priya: It means what it's always meant: anything you post publicly can and will be used to create value for someone else. The curtain was just pulled back a little too far for their comfort.
Theo: So the fight over our data isn't over, they've just redrafted the battle lines.
Priya: Exactly. They just put up a 'Nothing to see here' sign.
Priya: Alright, let's talk about the AI model wars. Alliances are forming, and Replit—the huge online coding platform—just switched teams.
Theo: Ooh, drama! They were working with Anthropic, right? Using their Claude model.
Priya: 'Were' is the operative word. Over the weekend, Replit's CEO Amjad Masad announced that OpenAI is now their default model provider.
Theo: Default! That's a huge deal. Most people stick with the default. This is a massive win for OpenAI.
Priya: And a major snub for Anthropic. The CEO says Claude will still be available, but it's been demoted. It’s no longer the first tool you’re handed.
Theo: So millions of developers are now funneled towards OpenAI's tech. What does this mean for the average person?
Priya: It means the platform wars are trickling down. The tools that build the apps you use are making bets. It's like if every construction company suddenly decided to use only one brand of hammer.
Theo: The other hammers are still on the shelf, but everyone's reaching for the same one first.
Priya: Precisely. It entrenches a market leader, making it harder for anyone else to get a foothold. For us, that could mean less diversity in AI experiences down the road.
Theo: So the battle for AI dominance is being fought in code editors and developer platforms—and OpenAI just captured a big piece of territory.
Theo: Alright, Priya, ChatGPT felt like it changed everything overnight. So where is my laundry-folding robot?
Priya: It's still a pile of very expensive parts in a lab somewhere. According to Amulya Vishwanath, a former Nvidia exec turned venture capitalist, we're not even close to a 'ChatGPT moment' for robotics.
Theo: She would know—she spent seven years on Nvidia's robotics platforms. So what's the holdup?
Priya: She calls this the 'Lego era' of robotics. We have all the basic building blocks—actuators, sensors, arms—but we don't have the intelligent 'glue' to make them work together in the messy real world.
Theo: So we have the body parts, but not the brain.
Priya: Exactly. A chatbot failing is just bad text. A robot failing means it dropped a priceless vase or drove into a wall. The real world has physics, and physics is a much harder problem to solve than language.
Theo: That makes sense. It's one thing to generate a sonnet, it's another to pick up an egg without breaking it.
Priya: Right. The 'ChatGPT moment' will come when a robot can learn a physical task as easily as a language model learns from text. That breakthrough, she says, is still a long way off.
Theo: Meaning you can probably keep your job for a while longer. The physical world is proving much harder for AI to conquer than the digital one.
Theo: Okay, let's recap.
Priya: Meta rewrote its terms so it sounds less like they're strip-mining your public data for AI, even though they say nothing has really changed.
Theo: Replit, a popular coding site, made OpenAI its default provider, giving it a huge leg up over rivals like Anthropic.
Priya: And robotics is still in its 'Lego era,' so your robot butler remains a distant dream.
Theo: Before we go, a little piece of financial history. The first ATM was installed in London in 1967. Its inventor, John Shepherd-Barron, came up with the four-digit PIN because that's all his wife could remember.
Priya: So we have one woman's memory to thank for decades of easily guessable PINs. 1-2-3-4. Groundbreaking.
Theo: Hey, a hero is a hero! She saved us all from 8-digit PINs. Think of her next time you're at the cash machine.
Theo: That's our show. Thanks for listening to Money Daily.
Priya: We'll be back tomorrow. And maybe think about a PIN that's longer than four digits.
This show is made with AI: the hosts’ voices are synthetic and the scripts are AI-assisted. Every story links to its original source.