9 min read
  • #conferences
  • #kawaiicon
  • #security-leadership

Kawaiicon and Jazz

My first Kawaiicon: router exploits, red teams, quantum maths over my head, and a jazz band in a Wellington bookshop that got me thinking about security teams.

On the flight down to Wellington for my first Kawaiicon, I had a few questions on my mind. How much of the conference would be about AI? My guess was like 50%. How much about quantum computing? Guessed maybe 1 talk, it’s still mostly science fiction, right?

And perhaps more importantly, was this conference really for someone like me? I’ve attended a few security conferences in New Zealand. NZ CISO was flashy, but a little sales-focused for my liking. OWASP was almost the opposite. A much smaller affair, seemingly held together by a handful of dedicated volunteers who clearly care deeply about what they’re doing.

My background is infrastructure and cloud, and having moved further into security architecture and leadership rather than hands-on engineering. My engineering skills and deep tech understanding are pretty crappy these days. So I wasn’t entirely sure what to expect. Would Kawaiicon be two days of extraordinarily clever people getting excited about very specific exploits? Or would there be some broader security lessons in there?

Who knows. Let’s find out.

Day one

The opening keynote is a sitcom-style performance about vibe coding appearing in a classic dev team environment. It’s funny, slightly ridiculous, and uncomfortably familiar.

One developer is embracing AI and producing code at incredible speed. Another is being more cautious, following security processes and questioning what the machine is doing. Management, of course, loves the productivity numbers.

There’s hard-coded credentials, questionable security practices, code effectively marking its own homework, and the inevitable consequences of trusting systems we don’t properly understand.

Anyway, the careful engineer gets fired due to low Jira stats, classic. The young vibe-coding engineer doesn’t fare much better, but turns out it’s because someone else vibe coded an integration to close Jira tickets that wasn’t working, and leadership just took the numbers haha! And eventually an unregulated Chinese AI model with access to long-loved credentials from a company they had been fired from, manages to create a fairly spectacular mess, essentially collapsing the digital world.

Many things to take away from this weirdly plausible scenario.

  1. Revoke the keys when people leave the organisation
  2. People making restructuring decisions should listen to people closer to the ground, no shit, but sure it’s often missed.
  3. A drunk individual with some badly guard railed AI, and inadvertent access to unlimited tokens, can legit cause a lot of harm.

It’s an interesting way to kick off the conference, and I’m expecting AI to dominate the next two days. Except it doesn’t.

Instead, we’re quickly into the world of vulnerability research.

One of my highlights is a young independent researcher, MrBruh, talking about his work investigating consumer routers.

He’s only 22! genuinely passionate about what he is doing, and speaks his mind, which is hilarious, and has clearly spent an extraordinary amount of time understanding how these devices work and finding ways to exploit them.

His research uncovered 28 vulnerabilities across six major vendors, achieving remote code execution against products from each of them.

I’m fascinated by the technical achievement, but also by the person behind it. How the hell does someone spend six months researching something like this? What motivates them? Does he have a day job?

There’s something quite extraordinary about somebody taking a relatively simple question, following their curiosity, and eventually uncovering something the people who designed the technology hadn’t anticipated.

One thing I would have liked to understand better was the real-world impact of these vulnerabilities. I appreciate that a compromised router can provide a foothold into a network, but who’s actually exploiting these things, and for what purpose? Are we talking botnets, traffic interception, or something else entirely? I found the research fascinating, but I’d love to connect it back to the actual threat facing an average household.

A talk on incident management brings things back into familiar territory.

One point really resonates: human error is not a root cause, I like that.

People make mistakes, but they make them within systems, cultures, processes and incentives created by organisations.

If a business consistently prioritises production and delivery over safety, should we really be surprised when something goes wrong?

An unexpected bit of jazz

After the first day, I find myself looking for something to do in Wellington. I have to admit I wasn’t seeking out a continuation of the conference.

As I do with a lot of things now, I outsourced the options to ChatGPT… Which sends me to a live jazz show in a bookshop down the road from where I am staying. Not exactly what I expected from an evening at a cybersecurity conference, but a welcomed change of pace.

And somewhere between the books and the music, I start thinking about the people I’d been listening to all day.

Jazz is an interesting thing. As far as I am aware, It’s a musical tradition that’s been developing for a very very very long time.

The sax player was incredible. Clearly someone who had spent years mastering his craft, much like the young researcher I’d watched earlier that day who had spent months finding vulnerabilities in home routers.

But what really interested me was the band as a whole. Each musician was obviously extremely talented in their own right, but the performance wasn’t just a collection of impressive solos. They were listening to one another, responding, sometimes taking the lead and sometimes stepping back.

There were clearly rules and a shared understanding of what they were playing, but within that structure was an enormous amount of freedom to improvise. And that’s where I started thinking about my own role in security.

I’m not the person who’s going to spend six months reverse-engineering a router, just as I’m certainly not going to pick up a saxophone and join that band. But I spend a lot of time thinking about how to bring talented specialists together, give them direction without prescribing every step, and help turn their individual expertise into something that works as a whole.

It’s an interesting balance. Too much structure and you risk stifling the very creativity that makes these people exceptional. Too little, and you might have some extraordinary individual performances without ever really making any music.

It left me wondering whether structured, corporate approach to cybersecurity really gives people like that router researcher the freedom they need. How do we bring exceptional individual talent into an enterprise without losing some of what makes it exceptional?

Day two

Saturday brings another crazy unpredictable collection of subjects.

We start with red teaming, presented by a few wizards (Ita Halloween btw). Red teaming shouldn’t be about producing an entertaining story of how somebody broke into your environment. The real outcome is remediation.

Finding the vulnerability is one thing. Getting an organisation to understand it, prioritise it and actually change something is another.

And if you’re engaging a red team, why make their lives unnecessarily difficult? Give them context, documentation and access to the people they need. Then expect meaningful results in return.

That feels like a particularly important distinction for security leaders.

Next, a talk on DIY sovereign cloud questions our dependence on the big cloud providers and explores a co-operative approach to hosting infrastructure.

It’s an interesting reminder that just because the industry has converged on a handful of enormous providers doesn’t mean there aren’t other ways of doing things.

Then comes a presentation on the extraordinary global supply chains involved in manufacturing semiconductors.

It’s a reminder of just how many layers of abstraction exist beneath the technology we take for granted.

I spend much of my working life thinking about Azure, AWS, APIs and cloud architecture. It’s easy to forget that somewhere beneath a simple CLI command sits an almost unimaginably complicated network of mineral extraction, manufacturing, industrial machinery, international competition and geopolitics! These enormous industrial systems are described as Leviathans, which I googled during the talk, and are apparently mythical creatures? The name seems appropriate.

And while there’s relatively little discussion of AI directly, of course highly related. The enormous demand for compute, the concentration of manufacturing capability and the strategic competition surrounding the hardware that powers it.

Perhaps AI doesn’t have to be mentioned in every presentation to be influencing almost everything.

Later, a researcher takes us through an entertaining investigation into publicly exposed API keys, particularly those appearing on newly created, potentially vibe-coded websites.

The premise is wonderfully simple. If everybody is using AI to build websites and apps, surely some of them are accidentally publishing their API keys? So why not go looking?

The storytelling is excellent, and the underlying problem painfully simple. We keep introducing new ways to develop and deploy technology, but somehow the same old security mistakes keep coming along for the ride.

It also brings me back to the opening keynote. AI can dramatically lower the barrier to building software, but it doesn’t automatically teach someone how to build it securely.

Then there’s a talk about burnout and mental health. I have to admit, my immediate reaction was, what on earth is this doing at a cybersecurity conference?

But on reflection, perhaps that’s a bit shortsighted. We work in an industry where pressure, complexity and the expectation to constantly keep up are pretty much built into the job.

Maybe the fact I found the subject so out of place says more about my expectations of a security conference than it does about the talk itself.

Then quantum cryptography finally makes an appearance.

The mathematics quickly exceeds my ability to follow it, but at least I can get quantum off my mind, and I didn’t even get the punchline of that talk! Nothing registered at all. It was like a maths professor at a lecture in uni (or what I imagine that’s like….).

There are also talks about using LLMs to recreate old video games, and somebody enthusiastically making composite parts for a coffee machine and motorbike. At several points I find myself wondering what any of this has to do with cybersecurity.

But perhaps that’s part of the appeal.

Kawaiicon doesn’t always feel like a carefully curated corporate conference. It feels like a gathering of people who are genuinely interested in how things work, and who enjoy sharing that interest with others.

The flight home

Sitting at the airport, I’ve got quite a few things going through my head.

AI-assisted development is creating new variations of familiar problems, but the fundamentals haven’t gone anywhere: credentials need to be managed and revoked, findings need to lead to remediation, and security decisions still depend on understanding the systems and people behind them. And yes, a fair chunk of the quantum maths went completely over my head.

But I’m also thinking about the conference itself. I arrived wondering whether Kawaiicon was really for someone like me. I leave with a greater appreciation for the extraordinary technical craft on display, but also with a reminder that security is about more than individual technical brilliance.

Perhaps the challenge for those of us working in enterprise security is finding that balance between structure and freedom. Giving people the space to explore, challenge assumptions and occasionally go off-script, while still working towards something meaningful together.

A bit like that jazz band in the bookshop.