Get ready for a wave of AI hacking public services
The Big Tech Show
October 27, 2026
AI could soon make cyber-attacks on public services far more frequent, with Ireland potentially facing serious disruption to everything from healthcare to banking and transport. Adrian speaks to Peter Tanham, analyst at TheState.
Adrian Weckler
This is an Irish IndependentProduction.
Peter Tanham
The analogy I use is, sure, maybe the windows in my house, one or two might be open, but a burglar doesn't have time to come check every window on every house in the whole city. But now with AI, they do.
Adrian Weckler
You're listening to The Big Tech Show with Adrian Weckler, sponsored by Deal. Deal lets you hire, manage, pay and equip anyone, anywhere. Visit deel.com/bigtech.
Hello and you're welcome to The Big Tech Show with me, Adrian Weckler. Now, soon, any public service that can be hacked will be hacked because of AI. That's according to my guest today, Peter Tanham, former founder, CEO, political strategist, and recently departed Meta Global Data Policy Executive. You wrote a blog post this week predicting that by summer next year, at the latest, we'll see a slew of services under attack from AI bots. That sounds pretty terrible.
Peter Tanham
It is a scary prospect. Yeah. Thanks for having me on. I want to write this post because I was thinking, as you mentioned, coming out at the tech world, what is it that I see people in tech fully believing about the future that I don't think has got through to the rest of the world, to go to public sector, even small businesses or kind of medium-sized Irish businesses.
And I think the kind of scale and abilities of AI, especially around coding and especially around cyber abilities, I think are moving faster than a lot of people appreciate. So I have kind of three main ideas that I think, if you believe the blog post or the title of the blog post makes sense. The first is that AI is phenomenal at coding. So there are lots of headlines that claim AI is going to take all the lawyers jobs and the marketing jobs and everything. I don't think you need to believe that, right? Some of that is probably hype, some of that might be real. But what's definitely true is that they're phenomenal at coding. Most software developers don't write much code anymore. Most software companies have AI running all their code.
And that translates to cyber security. So they're brilliant and they're relentless at running cyber.
Adrian Weckler
So that means they're good at finding holes, vulnerabilities. Because one of the things you write, you say, all software, all computer systems in every organization are full of holes and bugs and vulnerabilities. And once we turn new advanced AI systems to the task of looking for these gaps, we find them everywhere. Tell us about Project Glasswing.
Peter Tanham
Exactly. So that's exactly the second thing I didn't think that most people appreciated is just how much software holds and vulnerabilities there are, both in the software and just in our practices. Like people don't update their machines and they haven't got their two-factor authentication switch on and all this kind of stuff.
But when we turn AI to the task of looking for these vulnerabilities, they find it. So Project Glasswing was a project by Anthropic. When they launched their Mythos model, which gave early access to this very advanced frontier coding model to a lot of trusted organizations. When they turned it at their own code base, they found just reams and reams of these. So Mozilla found, I think it was over 200 vulnerabilities. Now, 14-
Adrian Weckler
That was in Firefox.
Peter Tanham
In Firefox, exactly, in the browser.
Google Chrome, not with Anthropics, but with their own, found lots of vulnerabilities in Chrome. Just yesterday, there was a major vulnerability announced in PHP, which is a coding language that is used to run up to 70% of the world's websites. So a huge amount of government websites, Wikipedia, I know The Journal.E, a few others had to come offline while they patched this. So this all shows that whether we know it or not, a lot of us have been operating kind of a plan of security by obscurity, that there are all these holes. Like, we're all human, right? That we don't update our software. All the software is made by humans, and it has kind of holes and bugs in it. But the bad actors haven't had the time to probe and prod for these all.
So it is the analogy I use is, sure, maybe the windows in my house, one or two, might be open, but a burglar doesn't have time to come check every window in every house in the whole city. But now with AI, they do. They have robots that they can automate that can go.
Adrian Weckler
Yeah, if you had loads of robot burglar associates, you could swarm them all over the house, looking to see where the gaps were.
Read the other 21 minutes of this episode here.
$15 gets you this episode and 99 more. Credits never expire. After checkout, one link brings you back here to read it.
Not what you expected? Full refund within 14 days.
Thousands of transcripts fetched by people building searchable podcast archives
Fetch transcripts with code
The same credits work from a script or an AI agent. The demo key returns a sample episode in full, no card needed:
curl -H "x-api-key: pt_demo" \
https://spoken.md/transcripts/1000651996090Markdown with the speakers named, for your notes, your knowledge base, or anything that makes HTTP calls.
From $0.10 per transcript. No subscription. Credits never expire. Prices exclude VAT, added at checkout for EU customers. Not what you expected? Email us within 14 days with 20 or fewer credits used and we refund the pack in full.
Using your own key:
curl -H "x-api-key: YOUR_KEY" \
https://spoken.md/transcripts/1000798278676