Ivy: California just made it illegal for an algorithm to be your only boss.
Marcus: This is Marcus.
Ivy: And I'm Ivy.
Marcus: And you're listening to Broke It, your daily AI news for Friday, October 2nd, 2026. We've got a lot to get to.
Ivy: California just passed a slate of new AI laws for the workplace, putting some actual guardrails on automated surveillance and so-called 'robo-bosses'.
Marcus: We're also digging into CoreWeave's new platform, Forge. Is it the unified AI development layer they promise, or just more of the same?
Ivy: And get this: cybersecurity legend Kevin Mandia is back. His new startup, Armadin, just raised a quarter of a billion dollars to unleash AI agent swarms on enterprise security.
Marcus: Alright, let's start in California. Governor Newsom just signed a stack of bills that could totally reshape the future of work. What's the main takeaway, Ivy?
Ivy: The big one is AB 2930. It basically says an algorithm can't be the sole decision-maker for hiring, firing, discipline, or your pay.
Marcus: Okay, but the keyword there is 'sole'. So they can still use it, as long as a human is somewhere in the loop?
Ivy: Exactly. The law mandates 'human oversight'. But what constitutes meaningful oversight versus just rubber-stamping an algorithm's decision? The law is a little fuzzy there.
Marcus: But that's still a huge step! It means a person, not a program, is ultimately accountable. It also limits using AI to track workers off the clock or monitor what they say. For builders like me, that's a clear rulebook.
Ivy: A rulebook with some pretty big holes. A separate bill, AB 2671, requires employers to notify workers about any electronic monitoring. But that notification can be a one-time thing at hiring. Will anyone remember that two years later when their every keystroke is being logged?
Marcus: But they have to tell you! That's the point! It stops the secret surveillance. You know the terms when you take the job.
Ivy: And if you need the job, you have no choice but to accept the terms. It's a consent issue. The power dynamic hasn't changed, it's just been documented.
Marcus: I see it as progress. What does this mean for the average person working in California?
Ivy: It means if you get fired, you have the right to know if an algorithm was involved and you can appeal to a human. You can't be fired by an email from `noreply@skynet.com` with no recourse.
Marcus: [laughs] Right. And for companies, it means you need to audit your HR tech stack now and make sure you have clear human-in-the-loop processes. No more hiding behind the black box.
Ivy: We'll see how the courts define 'oversight'. I suspect that's where the real battles will be fought.
Marcus: Alright, let's shift gears to the developer side. CoreWeave, the company known for its specialized AI cloud, just launched a new product called Forge.
Ivy: They're pitching it as a 'unified development layer'—a one-stop-shop to connect your data, models, training, and deployment.
Marcus: Which sounds amazing! The MLOps landscape is so fragmented. You're constantly stitching together a dozen different services. A single pane of glass to manage everything would be a game-changer.
Ivy: Except it's not really a single pane of glass. It's more like... a glue factory.
Marcus: Ouch. What do you mean?
Ivy: From what I'm seeing, Forge isn't some new, standalone platform. It's an integration hub. It connects to the tools you already use—Hugging Face, Weights & Biases, you name it. It's not replacing the messy landscape, it's just putting a CoreWeave wrapper around it.
Marcus: But a good wrapper has value! If it handles the authentication and data flow between all those services, that alone saves developers a ton of headache. That's not nothing.
Ivy: It's not nothing, but it's not revolutionary. They're selling convenience, not innovation. And it creates another layer of lock-in to the CoreWeave ecosystem. I'm just not seeing the 'forge' so much as the 'façade'.
Marcus: [sighs] You're so cynical. So what does this mean for a developer?
Ivy: It means if you're already deep in the CoreWeave cloud, Forge might be a nice quality-of-life improvement. But if you're not, it's probably not a compelling reason to switch. It's a feature, not a moat.
Marcus: I still think a good glue factory can be a very valuable business. We'll have to see if developers agree.
Ivy: Okay, on to our next story. A name from cybersecurity's past is making a huge splash with a new venture: Kevin Mandia.
Marcus: The founder of Mandiant! A legend. What's he up to?
Ivy: His new startup is called Armadin. And it just raised two hundred and fifty-five... point five... million dollars.
Marcus: Wow— at what valuation?
Ivy: A two-point-five billion dollar valuation.
Marcus: [excited] Okay, that's serious money for a new company. What do they do that's worth billions before they've even shipped anything?
Ivy: They're building AI agent swarms for security. The idea is to unleash thousands of autonomous agents on a company's network to find vulnerabilities—like an automated, tireless red team.
Marcus: That is brilliant. Instead of a handful of human pentesters working for a week, you have a swarm of AI agents testing every possible attack vector, 24/7. That's a paradigm shift for security validation.
Ivy: In theory. But 'agent swarms' is one of those terms that sounds very cool and could mean anything. How do you coordinate them? How do you prevent them from accidentally taking down production systems? How do you distinguish a novel finding from noise? These are huge, unsolved problems.
Marcus: But with Mandia's reputation and a quarter-billion in the bank, he has the resources to solve them. This feels like the future of offensive security.
Ivy: It feels like a valuation based on a founder's track record and a hot buzzword. Not on proven technology. Call me when they publish peer-reviewed results showing it works better than a team of skilled humans.
Marcus: So what's the takeaway for a Chief Information Security Officer?
Ivy: The takeaway is: you're going to be hearing the 'AI agent swarm' pitch a lot in the coming year. Be prepared to ask very hard questions about the 'how'. Don't get blinded by the big names and big numbers.
Marcus: Alright, let's wrap the headlines.
Ivy: California's new rules for AI in the workplace are putting 'human oversight' on the map, taking on the robo-bosses.
Marcus: CoreWeave's new Forge platform is trying to unify MLOps, acting as a powerful glue for the tools developers already use.
Ivy: And cybersecurity icon Kevin Mandia's new agent-swarm startup, Armadin, raised a massive $255.5M at a $2.5B valuation, betting big on the future of automated security testing.
Marcus: And before we go... did you see that new text-to-video model from a research lab in Finland that specializes only in generating clips of cats falling off furniture?
Ivy: [dry] I did. The internet has finally achieved its ultimate form: an infinite supply of cats failing gracefully. A truly noble pursuit for our most advanced technologies.
Marcus: [laughs] The dataset is just called 'Feline Fails'. You can't make this stuff up. I think it's art.
Ivy: That's one word for it.
Marcus: Alright, that's our show! Thanks for listening to Broke It. We'll be back tomorrow with more AI news.
Ivy: Until then, try not to get fired by a spreadsheet. I'm Ivy.
Marcus: And I'm Marcus. See you next time.
This show is made with AI: the hosts’ voices are synthetic and the scripts are AI-assisted. Every story links to its original source.