Open Source Security Starts With the Language Layer
Open source security is getting harder to ignore. AI code generators pull in open source libraries and packages at a pace no team can review by hand. Vulnerabilities keep climbing, and attackers increasingly target package managers. Yet many organizations still treat open source as free software that simply appears when needed.
ActiveState CEO Abby Kearns examines that challenge with Alan Shimel at KubeCon 2026 in Salt Lake City. Kearns joined ActiveState in March after leading the Cloud Foundry Foundation and holding technology roles at Puppet.
History Is Repeating Itself
Kearns and Shimel both lived through the dot-com infrastructure buildout. Kearns recalls scrambling for servers and struggling to secure dark fiber for new data centers. She sees the same scaling limits returning with the AI boom.
She also compares today to the early cloud years. Teams used the cloud for years before Docker made containers easy for developers. That unlock started the cloud native wave. Generative AI has delivered a similar aha moment, but the real transformation is still ahead.
Org Change Is the Missing Catalyst
Individuals are becoming far more productive with AI. Most companies, however, are not yet seeing the same gains. Kearns says the unglamorous answer is organizational change management. Fast tools need supporting structures, clear goals and well-understood fundamentals. Without them, speed simply creates new bottlenecks.
Start Secure From the First Package
At ActiveState, Kearns is focused on a question many teams overlook. Is the language you build on actually secure? She says open source depends on real communities, contributors and maintainers. Organizations need to understand where their software comes from and how they maintain it.
ActiveState gives teams a trusted place to pull secure language libraries instead of grabbing packages straight from public registries. Its reseller partnership with Docker pairs those libraries with Docker Hardened Images. Kearns argues that fixing problems late in the pipeline is slow and expensive. Building open source security in from the start keeps teams moving faster.
Transcript
Hi, everyone. We're back here live at KubeCon, Salt Lake City. Yeah, I'll be honest, I think the crowd's died down.
They must be in sessions, or sometimes I notice after the afternoon snack, it dies down. I think people get a carb hangover or something. Maybe take a nap.
Yeah. Well, the snacks here are heavy carbs, I realized. They gave out those sugar donuts yesterday.
I don't know if you saw them. They gave sugar donuts, yes. " Those are- They're trying to give you that afternoon perk.
Yeah. Sometimes you need a little bit of a perk. It's a little...
But you know what goes up, comes down, and you crash. Well, you got to keep the perk up until you can crash. Absolutely, or until you get to your room, I guess.
Anyway, I want to introduce you to our next guest. Her name is Abby Kearns. Abby Kearns.
And Abby is the CEO of a company called ActiveState. Now, you may be saying ActiveState must be one of those new cloud-native startups. Yeah, maybe, except this is a startup that's been around for 30 years and still going strong.
Still going. So Abby, welcome to Techstrong TV. It's great to have you back on.
I am so excited to be back on with you. You're one of my favorite people, so it's always great to hang out. A pleasure.
A pleasure is mine. Before we talk about ActiveState, though, you and I were just talking, I gave Abby a copy of my book. We're going to take a picture after.
But in talking about it, you lived through that dotcom fiber network- All of it ... build out the same way I did. Give people a little of your history.
I started when we were quickly building out data centers and infrastructure to support this new thing called e-commerce. Yay. When the idea that we could actually not only have a website, but you could get money- Sell stuff ...
from the website. So all of a sudden, the big talk back then was that IT was all of a sudden going to move from a cost center to a revenue- To a revenue generator ... revenue generator.
And it was- Some things never change. I feel like we've been having that conversation ever since. Yeah.
But that was the really exciting moment, which is like, I can have a website, and all of a sudden it can bring in revenue, and it becomes a revenue stream. " Yeah. So not just can we sell tickets online, but can people book their tickets?
Can they view their miles? Check in. Yeah.
There wasn't quite check-in then. No. Check-in came a year later when we rolled out the first kiosk.
Ah. Which at the time were called self-service machines. Right.
We rolled out a kiosk, and it was a pretty crazy time because we're talking about taking people's credit cards and allowing people to log in, which was a new thing. And then more importantly, what did that mean behind the scenes? Because essentially we were automating people out of jobs, because at that time, American Airlines, when you did any of that activity, when someone wanted to know their miles or wanted to check in or wanted to- You called up ...
you called, and they had a- And it wasn't a chatbot either. No, there was a call center of lovely people that sat in their offices in their headquarters in Fort Worth, Texas. Yeah.
And they took those calls. And so there was a lot of consternation around the automation and the technology, and obviously we were doing data center buildouts like crazy. We could not get servers.
This just sounds so familiar. We could not get enough servers. A lot of the infrastructure ran on Sun- Yes ...
DL380s. We used UltraSPARC, but go ahead. Could not get Sun servers, so we were buying them off eBay.
You were finding them in basements and repurposing them. There was a lot of scrounging around, a lot of driving servers around to different data centers. Yep.
When we were building out the data centers, you couldn't get dark fiber. There was limitations on dark fiber access, and particularly if you wanted to run a highly resilient- Right ... data center, you needed two.
Which meant multiple carriers. You needed at least two carriers- Yeah ... for dark fiber when you could barely get- On different sides of the building.
I remember this ... you could barely get one. Exactly.
And so we ran into limitations on how many data centers you could spin up at a time, how fast you could scale. We ran into all the scaling limiters, and so yes, I do think that history is now slightly repeating itself. But you know what they say?
Those who don't learn the lessons from history are doomed to repeat it. Someone like you and I, who hopefully have learned our lessons, are not necessarily doomed to repeat the same mistakes. But I remember those days, my God, and that's what we got into the book talking about.
Yeah. So now we fast-forward. We're here.
It's happening again. I had this conversation with another person earlier today. I didn't think in my lifetime we would have that...
We've had cloud, we've had cloud native, we've had a lot of- Client server ... but nothing approaching what we saw in the dotcom. I didn't think we'd ever see that again.
I didn't either, but here we are, and I think- And here we are ... and I think generative AI has obviously changed it, much like we think about the early days of cloud native even. Cloud native, we talked about cloud and cloud native a little bit in 2007.
So AWS came online, and we started to talk about, oh, the cloud. We're going to be able to burst. We're going to be able to run hybrid clouds.
All of the potential that didn't really manifest because obviously the dependency was- The types of applications and how you architect today. And also, that shift and lift didn't work. You had to do- No ...
really a transformation, and that was hard. It was very hard, and I'd say between 2007 and 2013, we kind of like went around and around. 2013 is really the first time when Docker came out.
Yes. And Docker brought containers to the forefront. And if we're being honest, containers have been around since the '60s.
Mainframe. We had BSC Gel. I was using containers with Solaris zones in- Yes ...
the early 2000s. So containers weren't a new concept, but what Docker had done successfully is really translated to something easy to consume by developers. That was the key.
And that unlocked what became the cloud native wave, which is where we learned to write applications that could take advantage of the cloud. Yeah. So there was like a five, six year gap where people used the cloud, but they weren't really moving to it until that unlock happened.
And so that really changed the game, going back to that magic shift, and I think what we are seeing now is AI's been around since the '60s as well. Mm-hmm. But when we started getting these generative AI capabilities, the large language models, they were really simplified such that we were able to really understand the value deeply on an individual basis.
You could talk to it, right. You didn't- Yeah ... need a coding language or anything like this.
It just really simplified that experience, and all of a sudden the aha moment came, and I think we're seeing that unlock. And then what is the potential from that? Well, we can go anywhere.
Yeah. And we're still in that froth of the innovation. I think we're in that froth.
We are in that early cloud before cloud native. And going back to my book, this is exactly what we saw with the electric, right? By the time electricity came out, the Industrial Revolution was underway.
Steam engines, right, were the- Right ... running the day. Most factories had a central steam engine attached to belts and pulleys that operated all the machinery in the factory.
And when electricity and electric motors first came out, they very simply replaced that central steam engine with a central electric engine. But it was still attached to the belts and pulleys, and you got a 10, 15% bump- Right ... electric versus steam.
It wasn't until people like Henry Ford and some of the other innovators started putting electric engines- Right ... throughout the factory that you got a 4X, 5X. And the cloud to cloud native is that same hop.
Yes. Well, sort of. The cloud to cloud native made the build-out approachable.
It's scalable. Yeah. Well, just like approachable, because before that, going back to where we started in the early aughts when we were building out the servers, the data centers for e-commerce, your limit on scale was how fast you could buy the server, get it racked, cabled, configured, then deploy the application.
There was a lead time on any type of scale. Yes, there was. If you're optimistically a couple of weeks.
But depending on server availability, it could be a couple of months, and so that was your throttle, whereas cloud native basically took the throttle off. You could scale like that. Yeah.
I could burst quickly. I could all of a sudden spin up 100 containers in- No time ... no time.
And so basically, the throttle on our ability to scale, that lifted, and now we're really revisiting that again, is our ability to write software. If you are talented and you understand and you have the depth of expertise, it's unlimited. You can do 1,000X what you could do before.
Look, I didn't do code since Sun machines. I was the help support at my own hosting company. I did the graveyard shift.
So I had to learn Solaris and learn to program and so forth. I'm writing AppSec now. The question is, though, so okay, we got that, but what is that catalyst that's going to 4, 5X it?
Yes, I could be 1,000 times more productive, and we're seeing individuals be 4X, 5X more productive. Sure. But we're not really seeing companies yet for the most part.
It's not translating. We're missing that central electric engine to electric engines throughout the factory. We're missing that first generation cloud to cloud native container, Kubernetes type orchestration.
What's the light bulb for AI? What's that eureka? Well, unfortunately, it's the unsexy thing, which is org change management.
Like giving people tools that can allow them to run 1,000 times faster, then what? If you don't have the supporting structure, you don't have the supporting tools, it's just like the early days of cloud native. " Right?
Yeah. That was their big metric, and everyone's like, "I want to do that. " Like, great.
Here's the technology and the tools are there, but you're going to have to do all the hard work, which is to say the lead up to that. How do you build software? How are your teams structured?
How do you- That's really where they- No, the whole pipeline. The whole process. You're talking DevOps now, right?
Well, what DevOps aspired to be. Right. No, the old John Allcott, right?
10X a day. Sure. 10 releases a day.
People thought that was pipe dream. They were like, "What are you smoking? " And now with AI, people are doing what?
Thousands of releases a day. It's ridiculous. Yeah, it's nuts.
It goes back to the fundamentals, though. I think the first principles don't change, which is in order to ship software faster, you have to both have clarity of vision, understanding of the fundamentals, what are you actually building towards, what are the tools you're going to be using, and the structure. I think AI can allow us to do so much more, more quickly, but it really asks a lot of us to be very thoughtful about that.
What am I actually trying to do? What am I building towards? And to your point, yes, organizations, particularly large organizations, aren't seeing the benefits yet, but there's a lot of people work that has to happen.
Yeah. There's a lot. It's foundational stuff.
Abby, my friend, your marketing team is going to kill me if we don't say, okay, how does this relate to ActiveState? I guess we should talk about ActiveState. Yes.
A little bit. Well, I joined in March, and I was excited to come back to open source and security, and I think now is the moment where obviously all of these conversations are swirling, which is open source is more important than ever before. We're using more of it, going back to AI code generators.
What are they pulling in? They're pulling in open source language libraries, open source packages, et cetera. Security is becoming a hot topic again, which is great.
Yes, it is. Turns out running fast, you really want to make sure there's some- You need to be secure. You know- But we've also had our fingers burned a few times already, so.
Yeah. I think the number of vulnerabilities identified are growing. The number of exploitations through package managers is growing.
Obviously, this is becoming a pretty critical time, and so I really was excited to come to ActiveState to say, how can I be part of that? Which is to say, what's the one thing that's foundational to everything we're doing at any company? What language are you using?
Yes. Is that language secure? Oh, that probably should be, because if you're not, you're more exposed- Language barrier ...
than ever before. Yeah. But honestly, it's the one thing that people aren't paying that close attention to.
And that, again, is an old story. Yes. Open source has been around for decades.
It's grown, obviously, in popularity directly from upstream in the last 10 years, and I think people are taking open source for granted, obviously, and I think the acceptance that open source projects are kind of there as their free software, but it's not free. There's a community behind it. There's contributors, there's maintainers.
We're at KubeCon. There's a foundation behind a lot of these projects, and so it's really understanding and appreciating where does your software come from and how are you using that, and if you're using it, great, how are you maintaining that? And really, that awareness is step one, and then where we come in is just giving people a secure place to pull those packages from so that you're not just pulling them directly in from packages.
And that's what I wanted to hear. I had a company today, a startup, a real startup, just in the last couple of months, talking about what it would take to create truly a secure repo. And I said, "There are companies that are doing it.
They have secure packages, secure- Yeah ... " And that's really at the heart of what ActiveState is about, right? We're focused on the secure language libraries, and we're partnered with Docker now, who's really kind of digging in on, obviously, the hardened images side- Yeah ...
so that we can give customers that capability to say, "Hey, we want you to start secure. We'd like you to stay secure, too- Yeah ... " Start secure.
Because identifying these problems later, coming back and rebuilding- It's very- ... is very complicated, it's very hard, it's very expensive. It's much more expensive.
And frankly, going back to why organizations aren't moving as fast, if you're having to stop at the end of the pipeline and go back and rebuild, that's going to slow things down. 100%. How do people get started with ActiveState?
Super easy. It works right into your existing workflow. We integrate natively into your repository of choice, or you can just pip install us directly.
We don't ask you to change anything. com? Yes.
Is there a free trial, or... Free trial? Yes.
Check, check, check. Yes. You've got no excuse.
No excuse. You should try it. You should take open source security seriously.
If you have any questions, you can obviously reach out to me directly. I love it. Abby, thank you so much, and good luck.
Thank you. You know? Taking the reins here.
This is a company with history, but I think it's positioned well, and I think again, the ebbs of time are swirling around us, and it's a great opportunity to take what you've learned from before and make it count this time. I hope so. Well, we got the e-commerce thing figured out eventually.
We'll get this figured out, too. I think we will. I think we'll get there.
Abby Kearns, ActiveState. com, go check it out. Hey, I think we still have more to bring you today, so stay tuned.
We're going to be right back. You're watching Techstrong TV.