The Security Table · 2026-07-01 · 59 min
Key moments - from our scoring
Substance score
45 / 100
Five dimensions, 20 points each
The Security Table hosts Chris Romeo, Izar Tarandosh, and Matt Coles in a wide-ranging discussion centered on npm's upcoming security changes in version 12, which disable dangerous package installation behaviors by default rather than removing them entirely. npm will no longer execute pre-install, install, or post-install scripts from dependencies without explicit approval, block transitive git repository dependencies, and require explicit permission for remote URL installations - applying the principle of secure-by-default. The hosts debate whether this adequately addresses supply chain risks, particularly given that only 2% of npm packages actually require install scripts. They explore how AI coding tools might bypass these protections by re-enabling functionality, examine how other package managers like RPM handle this differently through signing and staged installation controls, and discuss the broader ecosystem challenge of extending such protections across 30,000+ package formats. The conversation touches on OpenSSF's code signing initiatives and whether cryptographic package provenance - modeled after Let's Encrypt's certificate automation - could become a commodity solution. While acknowledging npm's positive step, the speakers highlight that GitHub's 2020 acquisition of npm makes these changes particularly significant for the JavaScript ecosystem, and speculate whether the new defaults will actually break CI/CD pipelines relying on post-install scripts, potentially reversing the security gain through convenience.
npm will disable pre-install, install, and post-install scripts by default (requiring explicit approval), block transitive git repository dependencies, and require explicit permission for remote URL installations - applying secure-by-default principles without removing functionality entirely.
According to the episode's research, only 2% of npm packages need install scripts to function, suggesting most npm packages are libraries that work fine without them.
The Shai Hulud worm operated using post-install scripts to spread; npm's new changes would have prevented it from operating through npm, though it would have been forced to find alternative attack vectors through other package managers.
RPM allows developers to explicitly specify which installation stages (pre, install, post) to run; RPM packages also benefit from cryptographic signing and provenance verification, while npm relies more on implicit trust without formal signature verification.
Yes - if developers run npm with an 'approve-all' flag to unlock the build, the protection collapses entirely, and AI tools trained on pre-v12 npm behavior may attempt to re-enable dangerous functionality or find two-step process workarounds.
Our reviewer’s read on each dimension, with quotes from the episode.
The STRIDE and NPM discussions contain genuine practitioner insights - prompt injection as a vector not a threat, deterministic controls around non-deterministic AI, and the C undefined-behavior analogy - but they are buried under extended tangents about wheat color schemes, Emacs history, and Oregon Trail that consume a significant fraction of the 59 minutes. The insight-to-filler ratio is poor.
prompt injection is not a threat by itself...it's a vector. Uh, it's a way to get there. It's something that looks bad, but it's not by itself a threat because it's not doing things
the behavior of the AI may be non deterministic, but the controls that you put in it are deterministic. And those controls, they tie to specific scenarios of failure
The core contrarian thesis - that STRIDE is not broken, just misapplied, and that 'cross-context contamination' is just a renamed version of decades-old buffer-clearing hygiene - is a genuinely fresh and defensible take against the prevailing 'everything is new' narrative. The C undefined-behavior parallel is clever, but the episode doesn't develop these angles far enough to be truly standout.
how long have we known about the need to clear your buffers before you start a new session of anything that runs in a loop? How long have we known about, uh, mining swap files for passwords?
There are things in C that have undefined behaviors...you have. You have undefined defined behavior in the technology and you account for it. Right? That's stride.
Three co-hosts who are clearly working practitioners - one has contributed to the OWASP LLM Top 10 channel and has hands-on RPM/NPM experience, and the group demonstrates real depth on threat modeling - but there are no external guests, and the format limits the range of hard-earned practitioner knowledge on display.
I had the chance to go into the channel that's discussing prompt injection itself as a threat and put in there a lot of what I've been putting here on the podcast for a long time
databases have parole controls and have had them for what, 30 years? Because that stuff works.
A handful of concrete data points appear - the 2% npm figure, Let's Encrypt's 7 billion certificates, five named founding sponsors, STRIDE-LM, the Shai Hulud worm vector - but many are sourced from articles being read aloud rather than personal experience, and the STRIDE discussion stays largely abstract without named company examples or measured outcomes.
Only 2% of npm packages need install scripts to function
There were five, five original sponsors. Mozilla, eff, Cisco, Akamai, and Ident Trust
When the hosts engage the STRIDE article directly there is real back-and-forth and productive pushback - Chris challenges the healthcare example, Matt plays devil's advocate on likelihood vs. categorization - but the conversation repeatedly derails into unproductive banter (wheat, Emacs, Oregon Trail, GitHub acquisition confusion) that eats time without building on prior points.
Why wouldn't you consider AI in that threat model? Like, this is a strange example.
He had something, but he completely went off the rails.
Computed from the transcript - who did the talking, and the words that came up most.
In this episode, we dig into two things the security community loves to argue about: npm finally doing the right thing and whether STRIDE has any business being called dead. The npm v12 changes gate dangerous install script behavior by default, which is a good step forward and also about a decade overdue. Then we wade into a hot take claiming that STRIDE was built for a world that no longer exists, and we push back hard on the idea that non-deterministic AI systems need an entirely new threat-modeling religion rather than a better understanding of the one we already have. Also: wheat, Oregon Trail, and Emacs. Join the Conversation If your threat model failed because of an AI hallucination, was that STRIDE's fault or yours? FOLLOW OUR SOCIAL MEDIA: Twitter: @SecTablePodcast LinkedIn: The Security Table Podcast YouTube: The Security Table YouTube Channel Thanks for Listening!
Transcribed and scored by The B2B Podcast Index.
Speaker A: We're on Genie.
Speaker B: It's true. We've gone from Genie to Dano, Whatever that means.
Speaker C: Book them down.
Speaker B: That'll be the name of my. My seventh book. From Genie to Dano.
Speaker A: A History of Bookings.
Speaker B: Extreme Bookings. Well, it appears once again on the security table, you have entered a conversation that was already happening. But we want to welcome you, our trusted deterministic fans from around the.
Speaker A: Uh.
Speaker B: Would they be non deterministic?
Speaker C: Uh, they're almost certainly non deterministic.
Speaker A: No, they're deterministic because we can trust them.
Speaker B: True. I like that. Put that on the T shirt.
Speaker C: Wait, wait, wait. That's. Can you mathematically prove that?
Speaker A: Uh, I could call on people who actually know math to do it.
Speaker B: Yeah. I got the proof right here. One plus one equals two. Done.
Speaker A: On which base?
Speaker B: Proven. Seven.
Speaker A: It could be zero and carry the one.
Speaker B: True. I forgot about that. It's always the one that gets. Why, folks, I'm Chris Romeo, joined by Izar Tarandosh. Uh, Matt Coles. We are the Knights of the Security table. Actually, we're more like the knights that say knee than we are the Knights of the Roundtable. But okay, let's dive in here. Apparently, NPM has made some changes to Stuff and junk about dawn time.
Speaker A: Time is right.
Speaker B: So. Well, actually, it's going to be this month, apparently, that these changes will be happening about almost.
Speaker C: Darn.
Speaker B: Um, basically, they're taking out some bad behavior. If I had to summarize this, they're taking out some bad behavior that NPM could do. They're not removing it, they're just making it something that you have to enable. In other words, secure by defaulting it.
Speaker C: Enabling security by default. Yes, finally.
Speaker B: So just quickly to summarize what I see as the changes. Well, who am I kidding? What my AI says it sees as the changes. NPM install will not run pre install, install or post install scripts from dependencies unless explicitly approved, which is that secure by default. Then again, uh, NPM install will no longer fetch dependencies from git repos director transitive unless explicitly permitted. Removing code execution path where Git dependencies NPMRC file could alter which git executable is used. And then third, dependencies installed from remote URLs such as HTTP HTTPs tarballs would no longer be resolved unless explicitly permitted. So once again, does it sound like they're not. They're not taking anything out?
Speaker C: Nope. Because they don't want to break. They don't want to be break. Massive backward compatibility problems. Right?
Speaker B: Yeah. Which I think is the right thing to do for something with the scale
Speaker A: of npm, they are gating the bad behavior. Right.
Speaker C: Well, you have to make a decision. You as a developer. Well, or you uh, as a developer, or you as the AI who is taking the place of the developer, need to make a decision now. Interesting, actually. Interesting. Sorry. Now that brings up interesting thought. By leaving the behavior in but disabled by default, does that mean. Almost certainly any AI that has been trained on NPM prior to V12 will just re enable it and we're going to be. You're going to be broken again, or potentially broken again.
Speaker A: It will try to do it the wrong way. It will see that it doesn't work. It will try 30 other ways, including erasing your whole file system.
Speaker B: No, but it will work.
Speaker A: And then you are going to say, but wait, this change. And it's going to go back to you and say, you are absolutely right. I should have checked which version of NPM I was running before I destroyed the file system.
Speaker C: It's too late now.
Speaker B: I mean, so who needs a file system?
Speaker C: Well, because. Right, because V12, when V12 comes out, I mean this won't be a, this will be a problem in the first like couple weeks, but as soon as you get new models that are trained on V12, they'll know V12 has different behavior.
Speaker A: But to me, the really interesting thing here is that NPM has taken this high profile as a, uh, as a package distribution system, uh, as a package manager. Right.
Speaker B: And
Speaker A: it's not a very new thing. For a while it was, for a while it was limited more to people who dealt in a certain part of the space.
Speaker C: But people wrote JavaScript, people who.
Speaker A: I didn't want to go there. But you went there. Yeah.
Speaker C: So what else are you supposed to go there? No, no, no, no. I mean it's Node Package Manager, Node JS Package Manager.
Speaker A: It's just that kind of person, people that go, oh, that's just for front end. No, no, no, no, let's write the backend. Because that's a great idea.
Speaker C: Yeah. Because somebody Decided to write JavaScript to write backend server servers.
Speaker A: That's for a different episode. But the thing now is, I think
Speaker C: we've had this episode before, Deja Vu.
Speaker A: So uh, what I'm interested here is when will the other 30,000 different package package formats and package managers out there actually do the same thing? Right. Because this is just the right thing to do. We should be making difficult for people to do bad things even without knowing that it's a bad Thing, right? So if you really trust uh, the package that you're pulling, then sure, by, by all means allow dangerous whatever the flag is. And uh, if you don't then at least stop and think before you do it and uh, consider the, the, the consequences of the things that you're asking your shell to do for you.
Speaker B: So I just made the connection here that these NPM changes would have prevent if they they had been in existence. The Shai Hulud worm would have been not allowed, wouldn't have been able to operate. It was using a post install script.
Speaker C: It would have had to follow a different vector, right? It would have had to find a different package manager that had bad behavior. It wouldn't have been through npm, it would have been through something else.
Speaker A: It would have been slowed down because a lot of people would just look at the output and say oh, you just have to able to add a flag. Let's do it.
Speaker C: Well actually that's an interesting question. Is there a bypass here? So if you disable but keep all the functionality, could a package re enable it like as a two step process?
Speaker A: Any, anything that can call another thing would be able to re enable it. Right? So if you have a uh, bad script that, that calls a good script with bad flags, you're going to get a bad, bad uh, outcome. But the thing here is that at the end of the day who is this going to to actually serve the people who already had a minimum of uh, security awareness to know that this might be a problem. So unless it comes out with a big honking banner saying you're trying to do this thing and this thing may damage your then uh, I don't know if a lot of people are going to extract a lot of content uh, from this.
Speaker C: Well I mean honestly right now with everybody doing or not everybody doing vibe coding or spec driven development, developers will never see it.
Speaker A: Yeah but this is going to break a lot of uh, pipelines down there, right?
Speaker C: Because they rely on pre and post
Speaker A: install functionality on installing a lot of uh, these packages. And some of them do the post install thing. I'm not saying all of them, but some of them I'm sure do. So there may be some breakage in the horizon if this becomes the, the default.
Speaker C: Well why wouldn't you just ask your AI to fix it? Which as you said may result in the deletion of your fourth of your entire application, but at least initially,
Speaker A: which perhaps carries us into the next one.
Speaker C: Well before we get there. How did, how did RPM solve this? Right, so You. I think we were talking earlier. Rpm
Speaker A: on rpm, you can say explicitly which stages of the install you want to run. Right? If you don't say anything, then it's going to run the whole thing. The pre, the uh, install and the post. Or you could say just run this one or just run that one. The thing is that the uh, RPMs, perhaps because of their age, I think they do have a bit more of a security veneer around them. So you have uh, uh, they are signed. I'm not, I'm not familiar with npm. M to the point of saying that those packages can be signed. I think that it's more of a provenance thing. Could be wrong. But uh, RPM does have the whole signature verification ecosystem around.
Speaker C: But signing doesn't doesn't eliminate bad behavior.
Speaker A: No, it doesn't, but it gives you at least, uh, uh, again, that's why I said the security veneer. It gives you the provenance. And if your sources are trusted, then by the transitive property you're saying, okay, I trust them to run the pre, post and install scripts the same way that you trust them to run their binaries. Right.
Speaker C: Well, I guess that's today it's implicit trust. Right. Because nobody's checking the packages. I mean, if they did, uh, do you know anybody who, I mean, do you know of anybody who goes and looks at all of their JavaScript packages like explicitly. I want this one, I want this one, I want this one.
Speaker A: Of course.
Speaker C: Okay. Besides Chris.
Speaker A: No. I know people who will bring stuff in, scan stuff, put them in their own repository and direct all their developers to only use that replacement, that uh, repository. Right. Because at the scale and speed that these things go, be honest with you, I'm not expecting anybody to do that verification by hand. It's like, oh, new shiny thing, yay.
Speaker B: And there's tools to do that as well. You don't have to build that type of internal, uh, repository from scratch.
Speaker A: Yeah, true. Mhm. But hey, you know what? All in all, a good step forward.
Speaker B: Well, yeah, it is, but there are some bypasses to it. That's what I was looking at right now. M. Your AI can bypass it is actually true. Um, if developers just run NPM approved scripts all to unlock their build, the protection collapses entirely. And it says AI coding tools adding dependencies at speed may outpace human review, risking the allow less becoming a mere formality.
Speaker C: Mm. Well, so the alternative is they disable the functionality entirely, make it a backwards breaking chain, you know, making a breaking change and then. And then what? So now you need to have explicit trust of packages, right? That become. That defeats the purpose of NPM perhaps. Or has it become like PyPi?
Speaker B: No, I mean it's NPM's goal should be the inherent trust of all packages that you get from it. Now uh, that's an easy thing to say, not as easy of a thing to implement. But isn't that why they exist for the future at least is to provide trusted packages? Aren't they a trusted repository of packages?
Speaker A: But are uh, they trusted because they said that they are to be trusted or are they trusted because by use people decided to trust them?
Speaker C: I mean people trust pypi, right? Or not?
Speaker A: No, I mean did they wake up in the morning and said we are going to build a trusted uh, uh, environment or did they just say we are going to be the place where people will come to get their packages?
Speaker B: I think it's option two.
Speaker A: I don't know.
Speaker C: Yeah, speed over convenient convenience over anything else, right?
Speaker B: Yeah but in the second article that we're going to talk about here, there was mention made of OpenSSFs. Maybe it wasn't in the second article, but apparently uh, no, they have a, they have a signing project for package.
Speaker A: Yeah the cosign. And um, I forget that it's actually a very good, good project with a lot of very good people in it.
Speaker C: Is it? It's like um, let's encrypt.
Speaker B: That's what I started saying about five years ago is like somebody needs to build the let's encrypt certificate version for packages. Because think about do you don't even think about TLS certificates anymore for public sites. It's just a commodity. You don't even pay for. Like they just get generated by this, this service. If we could get to that point with, excuse me, with packages like that would be to be able to have that provenance, true provenance and cryptographically proven providence of a package.
Speaker C: Uh, interestingly though, so let's getting a little sidetracked but let's encrypt verifies that you have ownership of the site before it gives you a certificate. How would it do that for software packages
Speaker B: have to be a repo, have to be a repo thing or you do the same exact thing in a repo. You have to put a file in a repo like you would do in a DNS info record or something to prove ownership.
Speaker A: Mhm.
Speaker B: There, I solved it. Somebody can go build it now.
Speaker C: There you go. Somebody write a spec. Somebody write a spec and get Claude to write it for you.
Speaker B: Yeah, somebody vibe code it today and let's get this thing deployed tomorrow.
Speaker C: What's the cost of that? Actually, who pays the cost of the, uh, certificates?
Speaker B: I think it's a not for profit. That's uh, charity. A charitable organization, last I heard. I don't know if some big companies. I feel like some big companies seeded it with money, but I'm double checking just because with everything else in the world, I don't know anything anymore. I just type words into an AI bot and it tells me
Speaker C: because, uh, because AI has become a, has become a search engine.
Speaker B: Yeah, there were five, five original sponsors. Mozilla, eff, Cisco, Akamai, and Ident Trust. M. That kicked in a bunch of money, but now the Gates Foundation's on board.
Speaker C: Well, there you go. Well, so maybe we can get them to uh, to make code signing certs.
Speaker B: Call to action issued more than se. Let's. Encrypt has issued more than 7 billion certificates since launch.
Speaker C: And how many of those sites have served malware?
Speaker B: I mean, some, um, 30, 40%.
Speaker C: But uh, what's in. What's in your thread bottle?
Speaker A: We have to do that commercial. One of these things.
Speaker B: All right, well, let's talk about this other thing. I think we've talked about the npm.
Speaker C: Yeah, it was a good change. By the way, when did GitHub own npm?
Speaker A: When? What? What?
Speaker B: GitHub bought npm?
Speaker C: Well, okay, the. It's GitHub announces there's an npm change. Why did GitHub announce.
Speaker B: That's weird. I didn't think of that.
Speaker A: Because they're the ones who are going to break the most.
Speaker C: It's.
Speaker B: It's.
Speaker C: It's a GitHub. Yeah, GitHub blog about npm. Our next npm major version.
Speaker A: Huh? Huh?
Speaker C: When did GitHub own npm?
Speaker B: I'm going to click on the actual link up there.
Speaker A: Yeah, I did. And it has a label on top. Retired.
Speaker C: What's weird? I just list that too. What is that? What's retired npm? The announcement.
Speaker A: The announcement. What the. Uh, okay, now what?
Speaker B: What? Maybe we're not in favor of this now. Now we're confused.
Speaker A: I'm, uh, confused. GitHub Enterprise 7. Okay, I, I need something to drink.
Speaker C: Oh, and then. Oh, and then the previous, the previous one is Dependabot. Version updates now support the Dano ecosystem.
Speaker B: GitHub acquired npm20 okay, uh, we're a
Speaker C: little bit behind the news.
Speaker A: What?
Speaker C: Get up.
Speaker B: Um, my mind is just what, er, er, is so upset about. This. He just left the studio.
Speaker A: Left?
Speaker B: He just left. He just got up and walked away and said he needed a drink.
Speaker C: He's going to get something stronger.
Speaker B: Six o' clock in the morning.
Speaker C: He's getting a monster.
Speaker A: I, I went to check if uh, uh, GitHub by mistake had acquired itself.
Speaker C: Well, apparently GitHub has acquired npm in 2020.
Speaker B: Where were we?
Speaker A: But wait, wait, wait, wait. Was NPM a thing that could be acquired?
Speaker B: I didn't know that
Speaker A: that's how out of NPM I am. Actually, since we mentioned anything uh, uh, web coded, I want to turn to our listeners uh, and ask for a que. Ask for help. So npm, if you do NPM audit, it will look into something called a package. Lock into the directory that you are, go through all those packages and their dependencies and see if there is any um, outstanding security alerts for them or whatever. You can see these. I don't know. That's, that's all good and nice for the packages in the project that you're working on and their dependencies, but there's all kinds of tools that you, you install using NPM which could be dash, G, global or private. I haven't found yet a way to scan those for security issues. So if anybody knows of something like that, drop us a line, give us a comment, focus on LinkedIn or something. Because if there isn't one, my question is why that's no less important, right? I mean even our AI agents are written in distributed through uh, npm. So shouldn't be a matter of going and say oh, I remember that once I installed this thing, let me see if I can update it. There should be a.
Speaker C: Will this change break AI installs now? Uh, I don't know, probably they rely on these pre and post install scripts to uh, deploy. In fact they probably do because you deploy the bits and then you run the post install script to actually do the install.
Speaker A: Yep, I guess so.
Speaker C: That's an interesting, that's an interesting thing because if they make this the default behavior, but it breaks all the installs for convenience, they'll probably be disabled by default soon.
Speaker B: It said in the last, in the Last version of 11 npm 11 they made them all warnings so that people would be getting some visibility that things are about to break.
Speaker A: There's a bunch of stuff that you can add with NPM like skills, uh, and this and that and the other one that are not exactly binaries but they have post install scripts because they ask all kinds of questions on the install.
Speaker B: According to my research. Only 2% of npm packages need install scripts to function.
Speaker A: I had uh, like the question is which 2%? I mean that, that makes sense because a lot of them, most of them I'd say are probably libraries that come as dependencies. Right. So you need them around in order to be able to run something. But exactly what Matt is saying that this packages that now are part of our AI. I'm sorry, I can't call them agents, our AI, uh, clients. Right. Uh, those things do seem to need post install. I remember in over the weekend I installed something and it asked a bunch of questions afterwards. I'm guessing that that was a post install. If it wasn't, then there's another thing that we have to turn off. But uh, the, the derivative here for, for me is that if it becomes a learned behavior that you have to run it with that dash, whatever, uh, to get those things to work, then what have we achieved?
Speaker C: It's the security theater. The button. Single button click. Where's your. Where's your single button keyboard? There. That's basically what it is all over again.
Speaker A: Thanks for remember. Yeah.
Speaker C: So I'm looking up the discussion thread on this. So the functionality exists but it doesn't exist in.
Speaker A: Uh.
Speaker C: Question is uh, want some user to their discussion. Will node 26 ship with npm 12? And it's an ongoing discussion about which version of Node will ship with this thing by default.
Speaker A: I think we should invite somebody to talk to us about the NPM universe and make us more educated about it. Uh, me at least because I don't feel like I know enough of the details there.
Speaker C: I mean that'll be interesting. That'd be an interesting episode anyway to talk about. Not just as you mentioned, there's other things right. That rpm but also go right. There's other sort of language with built in package manager discussions to be had.
Speaker B: Right?
Speaker A: Because now they are environments. They're not languages. Once upon a while, once upon a time you had VI and a compiler and Life was good. And now you have uh.
Speaker B: Life was not good.
Speaker A: Hey, Life was great.
Speaker B: Okay. No Ide.
Speaker C: It was better.
Speaker B: Show me my errors.
Speaker C: It was better than having Ed. Do you remember Ed?
Speaker B: What was the line by line one.
Speaker C: Ed.
Speaker A: That was Ed. Was it? Yeah, yeah, it was vi.
Speaker B: Uh, VI was an improvement over Emacs, which is Isar's. Go to.
Speaker A: Hey.
Speaker B: Hey.
Speaker A: No, Nobody says one word about uh, Emacs.
Speaker B: See, I told you this is his.
Speaker A: I wonder if somebody already wrote a Lisp package that runs Claude inside of Emacs. Uh, uh, I'm sure they have.
Speaker C: If they haven't, you will have it by next week.
Speaker A: Uh, no, I don't. I don't touch Emux anymore.
Speaker C: You don't have to. I'm sorry to say, you don't have to. Claude. Will. Yeah, have your friend Claude do it.
Speaker A: Those were good times.
Speaker C: Those were good times.
Speaker B: Tell me to use Emacs. It'll be like, no, I can't do that.
Speaker A: No, no, no, no, seriously, Seriously. I remember Eliza. Do you guys remember Eliza?
Speaker C: Eliza. Yes, a little bit.
Speaker A: Yeah. Eliza running inside of Emux. That was one of like the first things that I actually discovered on Emux. And it led me, did let me through a very, very deep rabbit hole of learning. Uh,
Speaker C: when did you get colorization in your, in your editor?
Speaker A: It was around 91, 92. There was the, uh, what was the name of the package? Transmogrifier, Something like that that would give you syntax, uh, uh, color and emacs. And the fun thing was it took me a long time to get to uh, um, a color scheme that I could work with.
Speaker C: And there was no accessibility at that point.
Speaker A: No. And just last week, just last week I was trying to uh, um, find a color for each different agent that I was, sorry, client that I was running. And um, I was looking for wheat. Wheat was like the, the default of uh, XEMUX when it came out. And ah, I just couldn't find the right RGB for wheat. And it drove me nuts.
Speaker C: You know there's a limited set, right? You could test them all and then, I mean, it is mostly white, right? So it's, it's yellow.
Speaker A: That looks like work. We don't do work anymore.
Speaker C: You don't have to. That's what your clients slash agents are for.
Speaker A: Okay, so what did I do? I found.
Speaker C: You have an agent swarm. Test each of these, look for wheat.
Speaker A: No, no, no, what you do is I went up in the attic, I found an old laptop that I had that run Linux circa, I don't remember when. And it had in it like my DOT files from way back when and it had DOT emacs and dot X. Uh, uh, uh, what was it? Or something like that.
Speaker B: Rc.
Speaker A: No, no, no, no, no, no, no. The DOT X files.
Speaker C: All right, I have a surprise for you. Is there?
Speaker A: And in one of them I had dot foreground equals DRGB for wheat. So I was able to find that.
Speaker C: I, I have, I have something for you.
Speaker A: No, don't tell me that Gemini found it for you.
Speaker C: I did. Uh, not Gemini, but there Is an Emacs with GitHub pages for the Wheat.
Speaker A: Yeah. You know what? I had much more fun looking at my old laptop.
Speaker C: Including. Including MRE M. Including the RGB color, uh, blocks. So you know exactly in CSS so you can apply it to your HTML pages.
Speaker A: Yeah.
Speaker B: I heard wheat's making a comeback. It's supposed to be the color of the year for 2017.
Speaker A: It's easier on the eyes.
Speaker C: It is.
Speaker B: Everybody's going to be using it. It's going to be. Houses are going to be painted wheat.
Speaker A: Okay, so wheat is the new dark mode.
Speaker B: Yeah, it is. I heard that, too, from people. And people in Silicon Valley.
Speaker A: I heard that T shirt. Okay, cool.
Speaker B: I mean, they were on the TV show.
Speaker C: We. We done black.
Speaker A: What?
Speaker C: We done black, so.
Speaker A: Oh, we don't blank. So we're going to, uh, jump into. The STL is broken. Or we're going to keep it for next week.
Speaker C: I think we.
Speaker B: Oh, no, let's talk about the Stride built world longer exists. I kind of said that wrong.
Speaker C: Um, Stride was built for a world that no longer exists, is the topic. Yes.
Speaker A: In a world that doesn't exist anymore. A methodology.
Speaker B: Rose, you really need that guy. Once again, you need that guy from the movie previews. In a world that doesn't exist. Stride is missing it.
Speaker A: Hey, we are continuously searching for it.
Speaker C: It's a stride out of time.
Speaker B: All right, so let's do this.
Speaker A: Has anybody come up with a strip bear agent anymore already?
Speaker B: Probably.
Speaker C: I think you just invented it.
Speaker A: Uh, yeah, I think I'm gonna pull this Stripe for client stake on the ground for that one.
Speaker C: Stride for client.
Speaker A: Stripe for client. No, that's a consulting gig.
Speaker B: Uh,
Speaker C: so, uh, I mean, do we
Speaker B: want to be here? Do we want to be just normal mean levels, or do we want to.
Speaker A: No, no, Wait, wait, wait, wait. I think I'm in the wrong place. I mean, the, uh, STL is broken, and that's actually good news.
Speaker B: You're the wrong article. We're in. Stride was built for a world that no longer.
Speaker C: Or move backwards because it was posted.
Speaker A: Oh, backwards. Backwards. Okay, okay. Let me go look for it. Okay, you guys start. I'll, uh, uh, I'll drink it.
Speaker C: It's in chat, by the way.
Speaker B: Just watched. These are just catching up, too, because he was still.
Speaker C: He's still stuck on his wheat.
Speaker B: He's still. He's still just looking for the color wheat. So if, you know, he's telling.
Speaker C: No, he's playing. He's playing settlers like catan, and he's
Speaker A: just Just assume that I'm stuck. Like it's, it's. It's the safe choice.
Speaker B: Just, he's playing, uh, what's the, what's the game? That. Oregon Trail. Oregon Trail over there.
Speaker A: Oh, you just died of, uh. What was it? Dysentery?
Speaker B: I don't know. Izar died of dysentery again and again. All right, so hey, I played a
Speaker A: single on Apple too.
Speaker C: I didn't play it on Apple II in green. Uh, yep.
Speaker B: Anyway, all right, so how did I
Speaker A: end in a sentence?
Speaker B: Going to be normal here when we respond to this.
Speaker C: So two things about stride. Stride is both a, uh, threat categorization and a, uh, methodology. Yeah, right. So you have stride, the spoofing, tampering, repudiation, et cetera. And you have per. Element and per interaction.
Speaker B: Okay, agreed.
Speaker A: So far, so good.
Speaker C: Both of those, I would argue, still apply in an AI world. And in fact, it applies, I think, even more so in an AI world. Because what's an AI? An AI is a process. Uh, AI is a behavior has behaviors. And the stride threat categories are all about behaviors.
Speaker A: Oh, God, I'm going to be so mean here.
Speaker C: To who? To who are you going to be mean to whom? To whom?
Speaker A: To a world that no longer exists. I mean, why do people feel this, this. This need to put down something that has, uh, been serving us quite well? Might say not. Not exactly my choice of where to go, but as its uses, definitely. Right, sure. And. And then start saying, oh, no, no, no, doesn't work anymore. No, no, no, no. The word guys, for.
Speaker B: For.
Speaker A: For. For the love of Tulu. It's bytes. They go into cpu, they come out. We're not quantum image yet. It's not like all possible things happening will happen everywhere at the same time. No, it's bytes come from memory going to cpu. An opcode gets executed. A. It's the same thing as people saying, uh, you know what? We can't have maps anymore. No, no, cities can't have maps anymore because, hey, you know what cars can take anyway? Day one, right? You never know which way a car is going to take. Hey, it can drive this way, it can drive that way, it can drive the other way. So let's look. Maps anymore.
Speaker C: If you're in Boston, one way, streets don't matter. Right?
Speaker A: Well, that's a different thing. Jerusalem, same thing. I mean, you can comment in a different day and it's going completely different way. But my point is we are all running around saying, oh, all these things are non deterministic. Yeah, sure, fine. Yeah, yeah, yeah. You don't know exactly what's the next step, but the whole thing should be bound. Only a certain number of things can happen. You just don't know, okay, how it got there, the order.
Speaker B: But that's an important issue that. But I've even got something we can do before that. Can you, can you go to the top of this article and this example that he has here about the mid market healthcare company, Integrated ChatGPT. They ran a formal stride threat model and then the world fell apart.
Speaker A: Wait, I haven't seen the word falling apart yet. Okay.
Speaker B: Ah, well, so then week three, a patient called, their medical records appear in the ChatGPT response. Not the record, someone else's. Can't we tie each of these instances? Like summarize and then the next one was summarize. My previous chat history isn't tampering. In a refactored AI world, isn't this just tampering? And maybe stride needs to be. Some additional explanations need to be made for how stride applies to AI.
Speaker C: I keep seeing someone's ready to go explode over there.
Speaker B: Go ahead.
Speaker A: So the, the, the thing here is that. But what's not being said here, right, is that it's the same thing as, okay, coding agents. We already agreed that we should relate to our code coming out from uh, a coding agent as code from a, uh, junior developer, right? Check it, read it, review it, whatever. Just don't accept it at face value. On the other hand, when we're talking about chatbots that could have like a, uh, uh, query functionality, people are willing to accept those things as if they are like master librarians that know exactly where to go and get only the right thing and bring it right in front of you and put it in front of you. You know how you don't get things that can do what's being described here? By binding the amount of agency that they can have by using controls beyond the AI that will say, you know what, mine that record, but not that one. MHM databases have parole controls and have had them for what, 30 years? Because that stuff works. But just to say, hey AI, whenever somebody asks for a record, I want to show their records, nobody else's. Okay, that's not non determinism. That's ridiculous optimism. That's not how things work. So yeah, the behavior of the AI may be non deterministic, but the controls that you put in it are deterministic. And those controls, they tie to specific scenarios of failure. Those scenarios of failure tie to impact. Stride likes impact you probably, as you said, could bind those things back to stride. And it's not like all of a sudden, there is this new class of stuff that stride never saw before. And, oh, my God, we need another economy. We need layers. We need layers.
Speaker C: All right, so, uh, I'll play devil's advocate a little bit here. Why did stride LM exist? It's a rhetorical question. You don't have to answer.
Speaker A: Wait, wait, wait. Do we want to tell people we're
Speaker C: not familiar with, uh, stride LM right stride, plus lateral movement. And that's the plus lateral movement. That is the important part. Somebody decided that stride was necessary, but it was insufficient or it was lacking something. And so they added lateral movement. Right. And so you had the base, and then you extended it to fit a particular use case.
Speaker A: But that's the. That. That's the thing. What did you extend it with? You didn't extend it with one more impact. You stand. You extended it with one more way of using the previous five impacts to get to where you want.
Speaker B: True.
Speaker A: Right, right, right.
Speaker C: Utilize tampering to get a pivot.
Speaker A: Exactly. So it's the same thing as prompt injection. Right. The folks at OWASP who are doing the whole LLM, uh, top ten again, I had the chance to go into the channel that's discussing prompt injection itself as a threat and put in there a lot of what I've been putting here on the podcast for a long time that it's not a threat. Prompt injection is not a threat by itself. And I was expecting some pushback, or I was expecting to learn something that would completely change my view. And I did learn a lot of stuff for sure, but nothing that disagreed with me. And they were extremely gracious in allowing the newbie to step in, into something that clearly had been discussed ad nauseam over there. Right. And they said, yeah, it's not really a threat. It isn't. It's a vector. Uh, it's a way to get there. It's something that looks bad, but it's not by itself a threat because it's not doing things. You need other stuff on the tail end to actually have an impact on something. Right. So the element here, the lateral movement by itself, it wouldn't be a vector, it wouldn't be an impact. It would be something that takes you further. It would be even a result. You want to get that foothold so that you can get lateral movement so that you can go somewhere else and have more impact.
Speaker C: And you're touching upon something that's very important. Actually, it's mentioned in the next paragraph. It said the threat model said this was unlikely. This was the hallucination part, but the attacker made it repeatable. Stride doesn't decide likelihood. Stride is a category of threats. True or stride is a way of identifying threats based on a dfd. It doesn't decide likelihood. Something else decides likelihood. Traditionally, that's the humans who decide likelihood. They look at us at a particular threat and say, how is it likely to occur? Is it likely to occur? Is it something we should do something about? Right. To some extent, or actually, in reality, you're actually asking one of the four questions of the four question framework when you're applying stride. What can go wrong? And. Or what are we going to do about it? And one of those answers for what are we going to do about it May be nothing, because it's unlikely to occur. So it's not stride. It's a threat modeling process. And the decisions that are made that decide that that threat may be unlikely. And so if the threat model decides something's unlikely because of past behavior, but now you have repeatable behavior that says otherwise, update your threat model.
Speaker B: Um, I'm looking at this going, the people that should be fired are, uh, the people that didn't do a proper threat model, considering all of their threats like they ran a formal stride. Threat model is what it says at the top. Why wouldn't you consider AI in that threat model? Like, this is a strange example.
Speaker C: If you don't know that hallucinations can occur with cross conversation, well, then you
Speaker B: shouldn't have deployed a ChatGPT connected API into an application. Have the knowledge to understand how it worked. Another one.
Speaker A: How it worked.
Speaker C: Can you hallucinate? No, of course I can't hallucinate.
Speaker A: Wait, wait, wait, wait. When, when somebody says something like an hallucination occurred because of a, uh, past conversation.
Speaker C: Cross context.
Speaker A: Cross context.
Speaker C: Which we know. So maybe we know that now, maybe we didn't know that.
Speaker A: No, no, no, no, no, no, no, no, no. Let's. Let's go back there. Let's go back there. Let's forget the hallucination for a second. Let's talk about the cross context here. How long have we known about the need to clear your buffers before you start a new session of anything that runs in a loop? How long have we known about, uh, mining swap files for passwords? How long have we known about looking for data structures in a process memory to figure out what was happening in there? It's not a new thing, people.
Speaker C: How many People understand that NPM automatically does things when you install packages.
Speaker A: But that's the thing, we are talking. No, no, no.
Speaker C: We've known about it for decades. Nobody's paying attention to it. That doesn't make stride invalid.
Speaker A: You know what? No, no, no, that's for sure. That's for sure. That's not what I'm saying.
Speaker C: That's not what I'm saying again, just bring it back.
Speaker A: I agree with you completely. What I'm saying again is we are making names for this. Esoteric things that happen inside AIs, like, I don't know, cross context contamination. Oh my God. Ccc. Oh, it's a new Aphrodite. Looks good, but damn it, it's just applied stupidity. I'm starting a new session. I clear the memory.
Speaker B: Seems to be basic practice and good hygiene.
Speaker C: Well, actually, uh, it is good hygiene, but doesn't it happen for you all the time? And I'm not defending it, by the way. I'm just saying if you start a new application, you make an assumption. We've talked about assumptions a lot in threat modeling, right? That you're making an assumption that memory gets cleared for you by the compiler. You know, the compiler set up, you know, pre prefix code to take care
Speaker A: of what professionals call a wrong assumption.
Speaker C: As a computer science. As a computer scientist, what do you say?
Speaker B: Wait, are we professionals?
Speaker A: No, no, no, no, no. Sorry, sorry. As a computer scientist, I would expect somebody to make that wrong assumption. As somebody who actually wrote software for a living, I would say.
Speaker B: Are we in the professional category though?
Speaker C: Not at all.
Speaker B: Three of us.
Speaker A: Wait, what profession is it?
Speaker B: I mean, what profession?
Speaker A: Uh, no, I mean.
Speaker B: Casino gambler.
Speaker A: No, that's when we're talking about the probability or um, crypto. The thing here is. The thing here is we. When I say we, it's not even the royal we, because I don't think that we are. Because I like to think that we are looking at those things in a different light.
Speaker C: We versus we.
Speaker A: Yeah, right. The vertical we as opposed to the horizontal we. There is this rush to create. There's this whole.
Speaker B: I'm really confused right now.
Speaker C: I'm not a professional little Steve Martin there, were you?
Speaker A: Yeah, There is this whole rush, uh, to create a new horizon of thing. A new taxonomy, a new ontology, a new something of things. Right? And uh, and uh.
Speaker B: Oh, wait, hold on, I gotta steal third. He's actually third.
Speaker A: Uh, so, uh, there's this whole let's invent this whole thing. And, and I think that coming and saying that Something like Stride all of a sudden lost its meaning or it's not relevant anymore. Is it's such a leap in trying to coin a new revolution where there isn't one. No.
Speaker B: Let me give you an illustration. Let me give you an illustration that might, might help to wrap this, this thing up and put this article to bed. So remember the Model T? Right. Ford created this very early model car. I can look at the Model T and say, I can't believe it doesn't have airbags. How do they, you know it's not a car. Or seat belts. It's not a car because it doesn't have airbags and seat belts. Yes, it's still a car. It's still function. You can still drive it down the road. Have, have other things. Have cars progressed in over the years since the Model T? Yes, of course.
Speaker C: Are they still cars?
Speaker B: That's what we're talking about here. There are newer things. Yes. Is Stride the latest and greatest thing? No, there are things. There are, there are new method, not methodologies, but, um, mnemonics and things that have been created for threat modeling of AI.
Speaker C: Yep.
Speaker B: And so why would I try to drive a Model T across the country and expect it to be as safe as a modern day car?
Speaker A: I wouldn't.
Speaker B: I would look for a modern day car and I would leave my Model T in the museum. So I just suggested Strike needs to go in a museum. Dang it.
Speaker A: No, no, no, no, no, no. You got something here. You got something.
Speaker C: He had something, but he completely went off the rails.
Speaker B: He threw it in the museum.
Speaker A: He tried to go through the whole country. But the 4T is going to take you from point A to point B. Okay. Same way that a Ferrari is going to take. Probably it's going to be slower. Depends on the angle of the.
Speaker B: Oh, no, it is going to be slower.
Speaker A: You have to remember that for all the lack of airbags, for the fact that the brakes were made of wood, for the. They were. Actually my family had a Ford T for a while, but, uh, uh, 1929. Not a, not a 1910. But, um, for all the things that today we look at them and go, what? You know what, the brakes still stop the car, the engine still pulls the thing, the wheel still turns the car around. You are probably going to have to drive differently. And so I'm going to say, you know what, instead of a, uh, Ferrari, let's talk about, uh, a Model T and a Tesla. The Tesla is going to do a lot of stuff for you, but it's going to do the same things that the Model T did. Because what you're looking for is the base functionality. You want to get from point A
Speaker C: to point B. I'm going to say the car analogy is the wrong analogy here. Probably it's the motor vehicle.
Speaker B: We just spent five minutes on it.
Speaker C: You did, but I think you're going the wrong way.
Speaker B: I think you're nice.
Speaker A: We didn't see the sign. We didn't see the sign.
Speaker C: Yeah, yeah. Well, road sign.
Speaker B: Road.
Speaker C: Road signs or, or motor vehicle driving laws and regulations and conventions are probably the better example. And I, and um, I'm, I'm touching upon that because as we look further down the article, I know we're, we're running short, uh, on time on this episode, but the. It talks later about how the threat model doesn't have a control to address, um, prompt injection. Well, that's great because that's not Stride's job.
Speaker B: Right.
Speaker C: Stride's job is to find the threat. Stride's job is not to define a control. The control comes from you identify a threat and then you figure out what you're going to do about it.
Speaker B: Yeah, true.
Speaker C: And so you have conventions and you have standards and you have guidelines about how do you deal with certain threats. Stride doesn't tell you that. That's, that's one of my, that's one of my gripes about Stride is it expects an, uh, understanding of, um, a lot of knowledge around the threats to be able to address them.
Speaker A: That's the whole think like a hacker thing.
Speaker C: That, and if you, and if you don't understand, but also think like a computer scientist or think like a developer or think like somebody who knows how to read API standards, uh, API specifications. You have to understand the technology you're using in order to address a threat against that technology. So if the AI model has unexpected output and you identified that in your threat model, but you don't know how to fix that because you don't understand how the technology works. That's a different problem than stride is broken. That's. You as a technology deployer don't understand the technology you're deploying.
Speaker A: The, the fact is that I think that we, we are fast approaching the rise of, uh, if it hasn't yet, and I just don't know about it, the rise of a new, uh, profession, uh, which is the AI psychologist.
Speaker B: Right.
Speaker A: Because. No, seriously, what, what Matt, just, just put down that, ah, you know, the LLM could give you, uh, I won't say bad, I won't say inexact. I would say a different kind of output than you were actually expecting. Now between you and what generated that output, there are a lot of boxes. Some of them are gray, some of them are white, some of them are black. The black ones. At some point somebody is going to have to rise up and say I understand what's happening there and I can explain to you what you have to do in order to get less of that. It's going to be an AI, uh,
Speaker C: psychologist, you know, is the, is the psychologist for the AI or for the person using the AI?
Speaker A: I uh, think that's for the person using it. We are on the, the psychiatrist step already.
Speaker B: But uh, so it's for the AI.
Speaker A: It's for the AI. I, I mean, interesting. It's what Matt's saying, you know, that something can go wrong but you don't know what to do about it. And all of a sudden the uh. When you had code to look at and you could exactly trace. And now I'm going to talk a bit against myself. When you had the code to trace and you could figure out what the path was, then you could go back on it and say, oh, here's the bug, I'm going to fix it. Now that you have the LLM and I'm not talking about what happens outside with the result of the LLM later on that, that, that part I have established my position quite forcibly, but inside the, the black box, the non explicability of what happened inside to get to that result. Right. That's the bit where you can't go back and find. Oh, oh, oh, this weight is the one that caused the distance between the thing and the other thing and that's what caused the output. You don't have that, that level of explicability.
Speaker B: Right?
Speaker C: We don't have, we don't have that. Right. We don't, we don't know how to do that yet.
Speaker A: Exactly.
Speaker C: So because it's non deterministic also and,
Speaker A: and it's complicated math but uh, yeah,
Speaker C: we don't know how many dimensions there were involved and number of parameters, all that stuff.
Speaker A: Right, exactly. And when we do we forget because it's a law. Right. But um, what I'm trying to say is that in between that what could go wrong and what can we do about it? If you're not able to trace back to the root cause of what went wrong, it's going to be very hard to say this is what I can do about it. So people come up with all kinds of fences and guardrails and whatnot that go around that black box bit and hope for good things.
Speaker C: Yeah, I have a thought exercise for listeners and for you guys as well, because, uh, you may have forgotten it when you use C or C.
Speaker A: Now, don't go there. C, C. Keep it C. Okay.
Speaker C: All right, I'll keep it C. There are things in C that have undefined behaviors.
Speaker A: Yes.
Speaker C: Did you throw away using C because of undefined behaviors?
Speaker A: No, we defined the standard.
Speaker C: And when you had an application that exhibited an undefined, uh. Or, sorry, when you had an application that used functionality that may have fallen into the bucket of may exhibit undefined behavior, what'd you do?
Speaker A: I called it elegant.
Speaker B: Elegant.
Speaker C: But this is. This is the exact same. This is a similar problem.
Speaker A: Exactly the same thing.
Speaker C: You have. You have undefined defined behavior in the technology and you account for it. Right? That's stride.
Speaker B: May.
Speaker C: Stride may point you in that direction, but stride will not tell you how to address that undefined behavior. Right. You just have to deal with it. And as an application developer in the past, you would have said, well, for X number of boundary cases, something bad is going to happen and we're going to six, seven, crash, right? We're just going to bail because there's nothing else we can do.
Speaker B: I thought you said six, seven, crash for a second
Speaker C: when.
Speaker B: But it was sig sev, Sig sev, sig sev.
Speaker A: You know, now. Now you're talking about levels of behavior. Right, sorry. Levels of bad behavior. And one thing is to have a, uh, memory, uh, exception and end up in a. In a sick cell. And another one is having your model say that, uh, all of a sudden, I don't know, something ridiculous that has been said lately. Um, so many options. So many options.
Speaker C: Mathematically proven. Uh, something.
Speaker A: Exactly. No, something as simple as. Yes. Uh, uh, I don't know, that group of people are responsible for all the problems in the world, right? So what went wrong here in one side is, hey, you try to write in a memory area that you don't have a bit that says that you can write. So I'm going to raise my hands up and bail out. The other one is, well, you know what? Uh, the single doctrine with a lot of texts and a lot of input from different social media, and it got the impression that this was a fact, but we're not quite sure because, you know, there was just this amount and that amount, but then it went through reinforcement learning, and the person who was reading the thing decided, hey, you know what? This is a fact. So, yeah, yeah, yeah, sure, why not?
Speaker B: Go ahead.
Speaker A: That's a good answer. And it got into the model somehow. So the, The. The. The factors that could influence that bad, uh, behavior, they are so dispersed. And there's. There's such a pipeline of them, and then there is the, um, unexplainability, um, and. And undiscipility. Uh, I can't with this.
Speaker B: Wow.
Speaker C: All right. You should probably end it there before Azar goes completely apoplectic.
Speaker A: I can't say that one too. Yeah.
Speaker B: Ah, that's one of the words I can say. So I just go with it. When I work, it always comes aplectic.
Speaker C: If you could spell it, that would be even better.
Speaker A: But.
Speaker B: Yeah.
Speaker A: So bottom line, I think that, you know, to go back and declare that. That the old ways of looking at what's out there are wrong. No, it's dead. The old way of looking at new things is dead. It's a bit premature. It may show a bit of lack of understanding of the old thing itself and the race to create new things just to, well, let's put it as
Speaker B: ulterior motives to pop on social media.
Speaker A: Yes, there's more to talk about in
Speaker C: this article, but let's leave it there. What?
Speaker A: What? No, no, no, no.
Speaker C: We'll talk about this article.
Speaker A: Not talking about this article. Definitely not. Definitely not. No, no, no, no, no, no. Please, please, please, don't. Don't misunderstand me. But other things that we have seen that run to, you know, create these new paradigms and new approaches, and at the end, they end up doing the same thing.
Speaker B: Oh, uh, people are trying to get famous. Yeah, but you don't get famous by just, you know, reciting an old thing.
Speaker A: True. But you can get famous by pointing out that, you know, you can look at something in a different way without dismissing everything that came before. That's.
Speaker B: That's a whole other episode right there. That is a bit of a phenomena. You've just. You've just hit on a phenomena of. That seems like it's in the last five years or so that to create something new, you have to break something else down. Is like a pathway that people are approaching fame and not the fame in the 1980s NBC television program. I knew it was in your head, so I had to draw it out.
Speaker A: I can't. I can't, I can't.
Speaker B: All right, with that, we can no longer be here, because we don't. We've already used our one 1980s television theme song that we could sing. And, uh, if we let another one sneak out the FCC will shut us down. So we have to.
Speaker C: It's. It's fair use at this point.
Speaker A: I like it. The way that you are saying means that anything automatic listening to this thing can actually recognize the song that I am singing, which I'm gonna take as a compliment.
Speaker B: Yeah, no, I think we knew what it was. It worked.
Speaker A: You baited at one. Come on.
Speaker B: Of course I did. It's what I do. Folks, thanks for hanging, uh, around talking about NPM changes, strides forward in the future and whatnot. When it comes to AI and what.
Speaker C: And Model Ts.
Speaker B: And Model Ts. Yes.
Speaker A: Where are we going? We don't need stride. Yeah, or, or brakes, wooden or not.
Speaker B: We do need to re refactor the flux capacitor though. And I'll le you with that. As we say, ado Ado. Uh, yes, we do.
Other episodes covering the same guests and topics, from across The B2B Podcast Index.
Ryan Lopopolo: OpenAI's Framework for Shipping Code at 70 PRs/WeekThe AI Native Dev · on GitHub99 / 100
Episode 62: Gene Kim on the Evolution of Product Operating Models in the World of AIMik + One · on GitHub96 / 100
This Company’s Compensation Cycle Takes Five Minutes with Rowan SavageFNDN Series · on GitHub95 / 100
Cloudsmith with Cian ButlerRust in Production · on PyPI95 / 100
The trust paradox: How attackers weaponize legitimate SaaS platformsTalos Takes · on GitHub94 / 100
Rebooting Enterprise AI with MCP and KubernetesPractical AI · on GitHub88 / 100Mostly Dead or Mostly Back: The Zombie Resurrection of DAST in an AI WorldRealists At The Table: How To See Through The HypeThe Agentic Access Problem: When AI Becomes Its Own AdministratorThe Tool Creep Problem: When More Security Means Less SecurityThe Human In The Loop Illusion: Why AI Approvals Are Failing Security