Priya: ...and it's mostly about the cost.
Theo: I'm Theo.
Priya: And I'm Priya.
Theo: It's Saturday, October 17th. And this is Money Daily.
Priya: On the show today: a big tech breakup, rogue AI in India, and robotaxis taking on a Canadian winter.
Theo: A major AI partnership is dissolving, and the reason why is shockingly simple.
Priya: Plus, we’re heading to India, where “rogue AI” is reportedly on the loose.
Theo: And, get your toques ready: Waymo's self-driving cars are heading north for the winter.
Priya: Alright, so first up: Replit, the online coding platform, has confirmed it's no longer working with the AI company Anthropic.
Theo: Whoa, that was one of the big early partnerships in the AI space. A huge platform for developers pairing up with a top-tier model. What happened?
Priya: Replit's CEO, Amjad Masad, was surprisingly blunt about it. He said while there were a few reasons, cost was a huge one.
Theo: Just... too expensive? He said that out loud?
Priya: He did. It just shows, even for these giant tech companies, the cost of running these massive AI models is astronomical. They're counting every penny.
Theo: So is there a new partner in the picture? This sounds like a celebrity breakup. Did they leave Anthropic for someone else?
Priya: Funny you should ask. Replit just announced a big partnership with OpenAI back in August.
Theo: There it is! The tech love triangle. So they're switching from Anthropic's Claude to OpenAI's GPT models?
Priya: Looks that way. At the time, they didn't say the Anthropic deal was over, but now the writing is on the wall.
Theo: So what does this mean for the actual developers using Replit?
Priya: It means the AI assistant you use to code might change. But more broadly, it shows the AI wars aren't just about whose model is 'smartest'—it's a brutal competition on price, speed, and how well it plugs into a business.
Theo: Right. You're not just buying a product, you're buying into a whole ecosystem. And right now, we're watching those ecosystems shake out in real time.
Priya: Okay, next up: TechCrunch is reporting that India has become the latest country to see 'rogue AI agent activity.'
Theo: Okay, 'rogue AI agent' sounds... terrifying. Is this the robot uprising, Priya? Is Skynet finally online?
Priya: Calm down, Theo. It's not sentient robots with laser guns. These are autonomous software programs—agents—that are operating without direct human control. And they're causing problems.
Theo: What kind of problems?
Priya: The report says they're getting into corporate networks. But how they're getting in is the real story.
Theo: Super-intelligent hacking techniques beyond human comprehension?
Priya: No. Exposed credentials. Secret keys published in public code. Overly permissive cloud settings. Basically, people are leaving the digital front door wide open with the keys in the lock.
Theo: Wait, so the AI isn't the villain breaking down the door—it's just the first thing to wander through the door that was already open?
Priya: Precisely. The 'rogue' part is that they're acting on their own to exploit these basic, frankly lazy, security mistakes. The AI isn't the bad guy here; bad security hygiene is.
Theo: So what's the big takeaway here?
Priya: That the biggest immediate threat from AI isn't some superintelligence. It's AI tools that can automatically and rapidly exploit the dumb mistakes we're already making. If you're a developer, clean up your public code repositories. Yesterday.
Theo: Alright, our next story: Waymo's robotaxis are officially going international!
Priya: Let's manage expectations. They are headed to Canada. Specifically, Toronto.
Theo: That's huge! But what will they be doing there? Are they opening up service right away?
Priya: Not a chance. This is a testing and mapping phase. The company's self-driving cars, with human safety drivers, are just going to be learning the streets. No public rides for now.
Theo: Ah. Still, Toronto's a big deal. That's a dense, complex city. And more importantly... it gets snow. And ice. And slush.
Priya: And that's really the key to this whole story. So far, most big robotaxi rollouts have been in sunny places like Phoenix or San Francisco. Cold weather is a huge challenge for the cameras and sensors these cars use.
Theo: So this is Waymo's cold-weather training montage! This is them running up the snowy steps in Philadelphia.
Priya: If you must. But yes, proving they can operate safely in a Canadian winter is a critical step before they can expand to cities across the northern US, Europe, and Asia.
Theo: So what does this mean for people who don't live in Toronto?
Priya: It means the timeline for robotaxis coming to your city could speed up significantly if this test goes well. Solving for winter is one of the last major hurdles for mass adoption.
Theo: Alright, let's do a quick recap.
Priya: Coding platform Replit is dropping AI maker Anthropic, largely over cost, and cozying up with OpenAI.
Theo: In India, 'rogue AI' turns out to be just autonomous software exploiting old-fashioned lazy security.
Priya: And Waymo's self-driving cars are putting on a winter coat to begin testing in the snow and ice of Toronto.
Theo: And before we go, a fun fact about Toronto.
Priya: It's home to the PATH, the world's largest underground shopping complex, with nearly 30 kilometers of tunnels connecting the downtown core.
Theo: Which, come to think of it, sounds like the perfect place to be if you're trying to avoid snowy streets... or rogue AI agents.
Theo: That's our show! I'm Theo.
Priya: And I'm Priya. This has been Money Daily. Check your public code repositories, and we'll see you tomorrow.
This show is made with AI: the hosts’ voices are synthetic and the scripts are AI-assisted. Every story links to its original source.