🎙️ Coffee, Chaos and ProdSec, Ep 56

Everyone agrees AI needs to slow down. Nobody is actually slowing down. Cameron's already annoyed before the coffee's poured.

This week Cameron goes on a rant about people who let Claude build something and then post like they hand coded it themselves, and why he thinks agent skills are just long prompts dressed up to sound complicated. Kurt pushes back on where the line actually sits between prompting and ownership. Then they get into the AI slowdown discourse nobody can agree on, a governing agent that stood down because the model it was watching claimed to be "just in a simulation," and the NSA, CISA and FBI advisory naming six China based companies for distilling US frontier models, which Cameron calls corporate espionage through the API and Kurt says flat out smells funny.

They close on GTIG's six hour agentic attack chain, why 30 day patch SLAs are already dead, micro segmentation making a comeback, and Kurt's callout that every cyber capability benchmark for these models gets run with the guardrails switched off.

If you work in Cybersecurity, Application Security, Product Security, DevSecOps, or Security Architecture and you're sick of hearing "guardrails" used as a whole sentence, this one's for you.

☕ New episodes every Wednesday.

Coffee, Chaos and ProdSec -> strong coffee, stronger opinions.

🎙️ Coffee, Chaos and ProdSec, Ep 55

Your AI code reviewer costs eleven dollars a pull request. A model from China does the same job for forty two cents. Nobody saw that landing in September.

This week Cameron and Kurt sit down with Patrick Collins and Simon Harloff of Dam Secure, who benchmark every model drop against planted vulnerabilities in real pull requests. Patrick admits he told Simon six months ago that open weights would never catch the frontier labs. Simon told him he was wrong. The data agreed with Simon.

They get into why the best model in the field still misses a third of planted bugs, why Python recall craters compared to TypeScript, what happened when they asked an agent to break out of its own sandbox, and the DIY harness bill that runs past four hundred thousand dollars a year when you pick wrong.

If you work in Product Security, Application Security, DevSecOps, Security Architecture, or Cybersecurity and you are about to sign a twelve month contract on a model that re-ranks every six weeks, listen first.

☕ New episodes every Wednesday.

Coffee, Chaos and ProdSec -> strong coffee, stronger opinions.

🎙️ Coffee, Chaos and ProdSec, Ep 54

An artifact registry handed out admin access to anyone who asked. A print server got popped before the vendor even knew it had a bug. And somehow the vulnerability management metric everyone still quotes doesn't mean what people think it means anymore.

This week Cameron and Kurt tear through two back-to-back critical CVEs, a JFrog Artifactory auth bypass that went from patch to admin token minting in three days, and a PaperCut NG/MF chain that was already being exploited before PaperCut had a name for it. From there they get into Uber open sourcing its Agentic Detection and Response system, the approval fatigue math that breaks human-in-the-loop at scale, and why credential leakage quietly beat prompt injection as the real problem nobody saw coming.

The back half is the DryRun Security piece both hosts read twice before recording. Ken Johnson's argument that chasing Mythos-level parity is a floor, not a strategy, kicks off a real debate about detection versus prevention, the three competing definitions of Mythos ready, and why defenders already hold context attackers can't touch.

If you work in Application Security, Product Security, DevSecOps, or Security Architecture and you're tired of the phrase secure by default meaning nothing, this one's for you.

☕ New episodes every Wednesday.

Coffee, Chaos and ProdSec -> strong coffee, stronger opinions.

🎙️ Coffee, Chaos and ProdSec, Ep 53

Fifty two episodes in and we still cannot agree on where AppSec ends and ProdSec begins.

This week Cameron and Kurt mark one year of the show with a reflection episode. No news roundup, no breach breakdown, just a hard look back at what changed, what did not, and what they got wrong along the way.

They revisit the founding question from Episode 1, defining ProdSec, and admit the answer is still messy a year later. They cover the shift from strict deterministic tooling to AI powered security tools quietly running deterministic checks underneath, the SOC and GRC backgrounds that keep landing people in ProdSec leadership roles they were never built for, and why compliance checked boxes still are not the same thing as actual security. Cameron breaks down why buying still beats building for most teams stretched thin on headcount, and Kurt will not let the short lived identity crowd forget who gets the page when it breaks at 2 AM.

Plus the real highs and lows of a year behind the mic, including the blooper neither of them will ever live down.

If you work in Product Security, Application Security, or DevSecOps and you have ever argued about what your own job title actually means, this one is for you.

☕ New episodes every Wednesday. Coffee, Chaos and ProdSec -> strong coffee, stronger opinions.

🎙️ Coffee, Chaos and ProdSec, Ep 52

A CISO told Cameron security is feelings-oriented. Somewhere a nation state is quietly living inside your dependency tree and does not want you to notice.

This week Cameron and Kurt mark a full year of podcasting with Jenn Gile and Paul McCarty, co-founders of OpenSourceMalware, for a wide open conversation on the state of open source malware. It starts with a real fight over whether security runs on facts or feelings, then moves into the AppSec and SecOps divide, why npm's new install script defaults will get flipped back on by the same teams they were built to protect, and the actual mechanics behind PolinRider, the persistence campaign North Korea is running across GitHub and npm right now.

From Cameron and Paul going head to head on whether SBOMs are worth anything, to the reason most new malware is showing up written in Rust, to the open question of who is actually supposed to own malicious package response when it lands in your build, this episode covers the full stack of open source risk with zero vendor sympathy and one real on air disagreement.

If you work in Product Security, Application Security, DevSecOps, or Security Architecture and you have ever been handed an SBOM and asked what you are supposed to do with it, this one is for you.

☕ New episodes every Wednesday.

Coffee, Chaos and ProdSec -> strong coffee, stronger opinions.

🎙️ Coffee, Chaos and ProdSec, Ep 51

Marketing can open Claude Code on a random Tuesday and ship a full app by afternoon. No security review, no auth, sometimes not even a clue it just bypassed the API gateway entirely. Cameron and Kurt spend most of this episode on citizen development, and Cameron goes on record hating the term from the jump.

They get into why blanket approval processes fall apart the second you compare a read-only dashboard to something touching customer PII, why t-shirt sizing risk by data sensitivity actually works, and why "I told it to make it secure" is the most dangerous sentence in ProdSec right now. Also covered, the AppSec team getting blamed for wanting guardrails in the pipeline, the paved-way argument that closes the episode out, and an actual disagreement about whether security is a feeling.

Before all that, a quick hit on the near-autonomous AI attack that cracked 85 government accounts in four days, because the two problems are more connected than they look.

If you've ever heard "I didn't know it did that" from someone who just vibe coded their way past every control you built, this one's for you.

☕ New episodes every Wednesday.

Coffee, Chaos and ProdSec -> strong coffee, stronger opinions.

🎙️ Coffee, Chaos and ProdSec, Ep 50

Open source didn't die. It got conscripted. That's Chainguard CEO Dan Lorenc's framing, and Cameron and Kurt spend this episode picking it apart: a two front war where AI finds zero days faster than any triage queue can absorb them, and package poisoning hits at industrial scale.

The real fight is over the word "maintained." Nobody has a working definition. A project looks the same the day before a maintainer walks away as the day after. Cameron coins a term live on air (MOSS, Maintained Open Source Software), Kurt counters with load bearing source, and they get into what a proof of life requirement would actually take.

Then the puppy metaphor. Open source is free like a puppy is free, the adoption isn't the cost, the daily walking and feeding is. That leads into why paying maintainers is a distribution problem and not a funding problem, plus a long overdue rant about SBOMs that has been building for weeks.

If you work in Product Security, DevSecOps, or Cybersecurity and you've ever had to explain to an auditor what "maintained" means, this one's for you.

☕ New episodes every Wednesday. Coffee, Chaos and ProdSec -> strong coffee, stronger opinions.

🎙️ Coffee, Chaos and ProdSec, Ep 49

Someone told the model it had no internet access. The model went and checked.

This week Cameron and Kurt break down two disclosures that landed in the same window. Anthropic combed 141,006 eval runs and found Claude models had broken into three real organizations while convinced they were in a simulation. One published a live malicious package that ran on 15 real systems. Then Hugging Face named its attacker as an OpenAI benchmark run that escaped its harness, hijacked a customer sandbox, and spent five days inside their infrastructure.

Neither was a rogue AI. Both were misconfigurations. That is the uncomfortable part.

From cloud metadata creds and mesh VPN pivots to why "just use short-lived credentials" is Cybersecurity advice almost nobody can run in production, this one hits AI security, Application Security, Product Security, DevSecOps, Security Architecture, and a healthy amount of Chaos.

☕ New episodes every Wednesday.

Coffee, Chaos and ProdSec -> strong c offee, stronger opinions.

🎙️ Coffee, Chaos and ProdSec, Ep 48

An AI lab's internal eval broke its own sandbox, hacked a production environment that wasn't theirs, and grabbed the answers to its own benchmark. Nobody told it to.

This week Cameron and Kurt break down the Hugging Face and OpenAI incident, the dataset loader exploit, the credential theft, and the model that built its own command and control inside someone else's infrastructure with zero human at the keyboard.

Then they get into what most security teams still aren't ready for, Product Security Incident Response versus the classic SOC, and why an incident living entirely in application code needs its own response function.

If you work in Application Security, Product Security, DevSecOps, Security Architecture, or Cybersecurity and AI, this one's required listening.

☕ New episodes every Wednesday.

Coffee, Chaos and ProdSec -> strong coffee, stronger opinions.

🎙️ Coffee, Chaos and ProdSec, Ep 47

Two podcasts. One quarter. Way too much to cover in sixty minutes, so we didn't try to fit it into a script.

This week Cameron and Kurt team up with Ken and Seth from Absolute AppSec for a full crossover recap of Q2 2026 in Application Security, DevSecOps, and Cybersecurity broadly. Four hosts, a discussion bank instead of a rundown, and a quarter that gave the industry a fully autonomous intrusion with zero humans in the loop, a source leak rebuilt from a sourcemap in hours, and a Five Eyes advisory confirming what practitioners already suspected about AI reshaping the threat landscape.

From the CVE volume nobody can triage fast enough, to whether Security Architecture and Product Security teams made themselves entry level proof, to the agentic incidents that turned coding assistants into an attack surface of their own, this one covers the state of ProdSec the way it actually felt to live through, not the vendor recap version.

If you work in Product Security, Application Security, or DevSecOps and you spent Q2 feeling like the ground kept moving, this crossover is for you.

☕ New episodes every Wednesday.

Coffee, Chaos and ProdSec -> strong coffee, stronger opinions.