Ivy: Over twelve hundred dollars... to fix a single bug.
Marcus: [beat] This is Marcus.
Ivy: And this is Ivy.
Marcus: And you're listening to Broke It. It's Friday, November 6th. We've got three big stories for you where AI and the real world collide.
Marcus: OpenAI built an AI to patch security bugs... and the price tag is staggering.
Ivy: Also, some of the biggest names in tech are suing California over a new law meant to protect teens online.
Marcus: And The New York Times might be dropping a few hundred million on your favorite movie app.
Marcus: So, OpenAI has created a new pipeline that can find and patch security vulnerabilities in Linux code, all on its own.
Ivy: Right, and these are the nasty kind of vulnerabilities—privilege escalation, the ones that let an attacker take over. The Hacker News reports this AI found and patched 948 of them.
Marcus: [excited] Nine hundred and forty-eight! That's incredible. It's like having a tireless army of white-hat hackers working 24/7.
Ivy: It is a big number. But let's talk about the cost, which is where my jaw hit the floor. As I said up top: over twelve hundred dollars... per bug.
Marcus: Whoa. Wait— twelve hundred dollars in compute costs?
Ivy: That's the number. In total, we're talking over 1.1 million dollars for this one project.
Marcus: Okay, that's a spicy meatball. But is it actually expensive? How much would it cost to have human engineers find and patch nearly a thousand critical bugs? I'd bet way more than a million dollars.
Ivy: Maybe. But a human can identify systemic issues. They understand context and can recommend architectural changes. Is this AI just playing the world's most expensive game of whack-a-mole?
Marcus: I see it as an investment. This is the first version! The cost will come down. And think of the potential damage if just one of those bugs got exploited. A million dollars is nothing compared to a major breach.
Ivy: Fair point. But what does this actually mean for people listening?
Marcus: It means the code that runs the internet is getting safer, faster. This is the beginning of an automated immune system for our entire digital world.
Ivy: [dry] For a price. And it means a new, very expensive front in the cybersecurity arms race has just opened up.
Ivy: Meanwhile, some of the biggest companies in the world are picking a fight with California. Meta, Google, and TikTok just sued the state's Attorney General.
Marcus: Over what? Let me guess, a privacy law?
Ivy: Close. It's a law that would restrict social media for kids under 16, requiring parental consent.
Marcus: And their argument for suing is... what? That's not a great look, suing to stop a law meant to protect kids.
Ivy: The classic one: free speech. TechCrunch reports their lawsuit claims the law is a 'blunt instrument' that unconstitutionally restricts access to information under the First Amendment.
Marcus: I can see that, though. The government telling a fifteen-year-old they can't browse TikTok feels like nanny-state overreach.
Ivy: Does it? Or is it like laws preventing kids from buying alcohol? There's a mountain of evidence these platforms are algorithmically optimized for addiction. They aren't a public library; they're a business model.
Marcus: But a blanket ban? What about kids who find supportive communities online because their home life is terrible? Or young activists organizing? This seems to cut them off at the knees.
Ivy: The platforms are framing this as an attack on free speech for teens, but let's be honest: it's an attack on their engagement model, which relies on maximizing time-on-site, regardless of age or mental health.
Marcus: So what's the ripple effect here? Why should people outside of California pay attention?
Ivy: This is the test case. If this law gets struck down, forget about any federal regulation for kids online. This fight is setting the precedent.
Marcus: Alright, let's talk about an acquisition that has film Twitter—or whatever we're calling it—buzzing. The New York Times is reportedly looking to buy Letterboxd.
Ivy: The movie logging app? Wow. For how much?
Marcus: TechCrunch is reporting a number around three hundred and thirty million dollars.
Ivy: Three hundred and thirty million? For what's essentially a shared diary for movie nerds? And I say that as a daily user, by the way.
Marcus: [laughs] Same here. But it makes perfect sense for the Times. They've been on a buying spree for community-based products: they bought The Athletic for sports, Wirecutter for reviews, not to mention their Games and Cooking apps...
Ivy: And Letterboxd has an extremely passionate user base. It's a firehose of data on cultural taste.
Marcus: Exactly! They can integrate it with their film reviews, use the data to inform their arts coverage, create new features... the potential is huge.
Ivy: Or... they could ruin it. The charm of Letterboxd is that it's this indie, nerdy corner of the internet. What happens when it's just another tab in the massive New York Times empire? Do they start plastering it with ads and paywalls?
Marcus: But they've been pretty hands-off with The Athletic, so I'm optimistic. Still, what does this mean if you actually use Letterboxd?
Ivy: It means your meticulously curated list of Japanese horror films from the 1960s is now a 330-million-dollar asset in the global media wars.
Marcus: And maybe you get a discount on a Times subscription with your Letterboxd Pro account!
Ivy: Or the price of Pro goes up to help pay for that massive price tag.
Marcus: Alright, let's do a quick recap.
Ivy: OpenAI has an AI that can patch Linux bugs, but it costs over a thousand dollars per fix.
Marcus: Meta, Google, and TikTok are suing California to block a law designed to protect teens online.
Ivy: And The New York Times might be buying the movie-logging app Letterboxd for a reported $330 million.
Marcus: Before we go... you know how AI image generators are famously terrible at drawing hands?
Ivy: The seven-fingered handshake. A modern classic of uncanny horror.
Marcus: Well, a new paper suggests it's a data problem. Most online photos show hands doing things, not just... being hands. The AI has almost no clean reference photos to learn from, so it's just guessing.
Ivy: [dry] So to fix AI, we all need to start posting more hand selfies? I don't think the world is ready for that.
Marcus: And that's our show! Thanks for listening to Broke It.
Ivy: We'll be back on Monday. Until then, maybe keep your hand pics to yourself.
This show is made with AI: the hosts’ voices are synthetic and the scripts are AI-assisted. Every story links to its original source.