Ivy: Twenty-five dollars. That's the going rate to have an AI agent hack a small business.
Marcus: This is Marcus.
Ivy: And I'm Ivy.
Marcus: It's Thursday, September 24th, and you're listening to Broke It, your daily AI news briefing.
Ivy: We've got three stories for you today, and they all feel... uncomfortably connected.
Marcus: First, we're talking about autonomous AI agents hacking online retailers for less than the price of a pizza.
Ivy: Then, Bill Gates has a new essay out, framing AI as a choice between utopia and... well, the other thing.
Marcus: And finally, the AI race just got a lot cheaper. OpenAI and Anthropic are in a price war, and things are heating up.
Marcus: Alright, let's start there: AI crime on a shoestring budget.
Ivy: A security firm, Gambit Security, just published a report showing how autonomous AI agents are successfully hacking e-commerce sites. Their estimate for a successful attack? Twenty-five dollars.
Marcus: Wait— twenty-five dollars? That's nothing. Is that just the compute cost to run the agent?
Ivy: That's the estimate for the whole shebang, end-to-end. This isn't a state-sponsored attack. It's a bored teenager with an API key and a grudge.
Marcus: Okay, that's terrifying. So the agent just... probes for vulnerabilities? Like a person would?
Ivy: Exactly. It automates the whole thing—recon, exploitation, all of it. Finds an old plugin, matches it to a known vulnerability, and builds the attack. For twenty. Five. Dollars.
Marcus: This is exactly what we've been saying. The double-edged sword of AI agents. It's incredible power... and that power can automate crime just as easily as anything else.
Ivy: Hold on. How reputable is 'Gambit Security'? Is this a real-world analysis or a proof-of-concept to sell their own services?
Marcus: That's a fair question. But even if it's just a proof-of-concept, it's a huge deal. The barrier to entry for cybercrime just hit rock bottom.
Ivy: [sighs] So if you run any kind of online store, your threat model just changed. You're not just worried about sophisticated hackers anymore. Now you've got to worry about cheap, relentless, automated AI agents.
Marcus: Which means get your security basics on lockdown. Patch everything. Now.
Ivy: Alright, let's pivot from the gritty reality of AI crime to the grand pronouncements of a billionaire. Bill Gates has some thoughts.
Marcus: He just dropped a new essay on his blog, Gates Notes, framing this moment as a critical crossroads for humanity.
Ivy: Ah yes, the crossroads. He argues the choices we make now will determine if AI becomes 'a great equalizer or a source of deep injustice'.
Marcus: But he's not wrong! Think of the good it could do. AI tutors for every kid on the planet, AI medical diagnostics in the middle of nowhere... that's the 'great equalizer' he's talking about. It's why I do what I do!
Ivy: [dry] We've heard this utopian vision for every major technology since the printing press. The internet was supposed to democratize information; instead we got misinformation at scale. Who, exactly, is making these 'choices' he's talking about?
Marcus: He's talking about all of us! Developers, policymakers, users—
Ivy: Is he? Or is he talking to a handful of CEOs at companies with enough GPUs to actually steer the ship? It's a lovely sentiment, but it feels disconnected from the power dynamics at play.
Marcus: You're being too cynical. The essay is a call to action. It's supposed to inspire us to build for the best outcome, to be intentional.
Ivy: And the 'deep injustice' part? He's right, that's the other path. Biased algorithms, mass surveillance, job displacement... it's not a long walk from here to there.
Marcus: Look, this debate is happening right now. And it's not just for tech billionaires. These are questions about the kind of society we all want to live in.
Ivy: It means pay attention to the policy, not just the products.
Marcus: Okay, let's bring this back down to earth. While Bill Gates is philosophizing, the actual AI race is just getting cheaper.
Ivy: A lot cheaper. In a classic tit-for-tat, Anthropic just dropped Claude Opus 5.5. Not to be outdone, OpenAI immediately announced two new, cheaper versions of GPT-6.
Marcus: [excited] And the price drops are insane. We're not talking 10 or 20 percent. Some of these new models cut the cost of high-end inference by 70 or 80 percent. EIGHTY percent, Ivy!
Ivy: Which just proves the real race isn't for some abstract 'AGI' finish line anymore. It's about price-to-performance. Who can deliver the most intelligence for the least money.
Marcus: This is a complete game-changer for developers. I had a project scoped out just two months ago that was way too expensive to build. With these new prices? Suddenly, it's on the table. This unlocks entire categories of apps that need that real-time smarts.
Ivy: Or it unlocks new categories of spam and garbage. Cheaper models just lower the barrier to entry for everything, good and bad. It makes high-volume, low-quality content generation practically free.
Marcus: But it also makes those AI tutors and medical tools we were just talking about more affordable! This is how the 'great equalizer' actually happens—by making the tech accessible to everyone.
Ivy: And it also makes the AI hacking agents from our first story cheaper. Remember that twenty-five-dollar price tag? Maybe next week it's five bucks. It all connects.
Marcus: Oof. Sobering thought. Okay, so what this all means is pretty simple. If you're a developer, your costs are coming down. If you're a user, get ready for more powerful AI features popping up everywhere.
Ivy: For better... and for worse.
Marcus: Alright, let's do a quick recap.
Ivy: AI agents can now hack a small business for just twenty-five bucks.
Marcus: Bill Gates says we're at a crossroads: build an AI utopia, or a dystopia.
Ivy: And powerful AI is getting way cheaper, which will probably fuel both of those outcomes.
Marcus: And before we go, some calendar trivia. Ivy, did you know today is National Punctuation Day?
Ivy: [sarcastic] Oh, is it. How wonderful. I'm sure the large language models are celebrating by generating reams of text with perfectly placed commas and zero hallucinations.
Marcus: [laughs] Hey, maybe the next big model breakthrough will be a true appreciation for the em-dash. A builder can dream.
Ivy: That's our show. You've been listening to Broke It.
Marcus: We'll be back tomorrow. Thanks for listening. And Ivy? Here's to the em-dash.
This show is made with AI: the hosts’ voices are synthetic and the scripts are AI-assisted. Every story links to its original source.