AI Bots Scam the Scammers: How Apate Hunts Fraud Before It Hits

SPEAKER_1 (0:09)

Imagine your phone rings, it's an unknown number, and, you know, against your better judgment, you actually pick it up.

SPEAKER_2 (0:15)

We have all been there. It is never good news.

SPEAKER_1 (0:17)

Right. And within seconds, this very urgent voice is telling you that your tax return is fraudulent, or your bank account is locked, and they need immediate payment.

SPEAKER_2 (0:27)

Cue the instant spike of anxiety.

SPEAKER_1 (0:29)

Exactly. But instead of feeling that panic and just hanging up, imagine you tap a button on your screen, and a customized AI persona takes over the call.

SPEAKER_2 (0:41)

Oh, I love where this is going.

SPEAKER_1 (0:43)

Yeah. And this AI sounds exactly like, say, a confused, slightly technologically illiterate grandfather.

And it just keeps the scammer on the line for 45 minutes.

SPEAKER_2 (0:55)

Just completely wasting their time.

SPEAKER_1 (0:57)

Wasting their time, frustrating them, to the point where they're literally screaming into the phone. But, and here's the kicker, while it's doing that, it is silently extracting the scammer's own crypto wallet address.

SPEAKER_2 (1:10)

That is just the ultimate reversal. Fighting fire with, well, a highly intelligent firewall.

SPEAKER_1 (1:15)

It really is. So today, we are doing a deep dive into an October 14, 2026 TechCrunch report by Ram Iyer.

SPEAKER_2 (1:22)

It's a fascinating piece of reporting.

SPEAKER_1 (1:24)

It really is. Our mission for this deep dive is to explore how an Australian startup named Apate.AI is sending artificial intelligence undercover to basically scam the scammers.

SPEAKER_2 (1:36)

Which is such a satisfying concept.

SPEAKER_1 (1:38)

Oh, absolutely. We're going to look at what intelligence they're actually gathering and the broader implications of fighting AI-empowered scams with even more AI.

SPEAKER_2 (1:47)

Because turning the tables on people who prey on our trust, it's a deeply satisfying premise. We all crave a way to fight back that goes beyond just tapping block collar.

SPEAKER_1 (1:58)

Or just ignoring every unknown number that pops up, which is what most of us do now.

SPEAKER_2 (2:02)

Right.

But as we look at the mechanics of what Apate is building here, the implications actually push way beyond just personal satisfaction. We are looking at a fundamental shift in digital security.

SPEAKER_1 (2:13)

Moving from defense to offense, right?

SPEAKER_2 (2:15)

Exactly. Moving away from passive firewalls and toward active offensive engagement.

SPEAKER_1 (2:21)

And the origin story of this whole operation really underscores that shift, because it started with a completely human reaction.

SPEAKER_2 (2:28)

The picnic story. The picnic story. I love this part.

SPEAKER_1 (2:31)

Yeah. So the founder and CEO of Apate, Dalli Kafar, was literally just at a family picnic when he got a scam call.

SPEAKER_2 (2:38)

As one does on a weekend.

SPEAKER_1 (2:39)

Right. But instead of ending the call, he decided to just play along. He played dumb, strung the scammer along for 44 minutes, entirely just to entertain his kids.

SPEAKER_2 (2:48)

Which, I mean, 44 minutes is an absolute eternity in the scammer ecosystem.

SPEAKER_1 (2:52)

I can't even imagine staying on the phone with anyone for 44 minutes, let alone a scammer.

SPEAKER_2 (2:56)

Well, right. But their entire business model is predicated on speed and volume. You know, burn through numbers, find a vulnerable target, extract the funds, move on.

SPEAKER_1 (3:04)

So slowing them down is basically stealing from them.

SPEAKER_2 (3:07)

Precisely. By keeping that person on the phone for nearly three quarters of an hour, Kafar forced this massive opportunity cost onto the fraudster.

SPEAKER_1 (3:15)

Because those were 44 minutes the criminal couldn't spend actually defrauding a real victim.

SPEAKER_2 (3:20)

Exactly. But, and this is the crucial part, Kafar also realized he had just wasted 44 minutes of his own Sunday.

SPEAKER_1 (3:27)

Right. Because you can't manually protect the world one phone call at a time. Nobody has the time for that.

SPEAKER_2 (3:31)

No, of course not. So the obvious question becomes whether a machine could automate that delay.

SPEAKER_1 (3:37)

And what I find so interesting about this origin is that the goal quickly evolved from just wasting their time to actual reconnaissance.

SPEAKER_2 (3:45)

It wasn't just about annoyance anymore.

SPEAKER_1 (3:47)

Exactly. While Kafar was at Macquarie University in Sydney, he and his team didn't just build a robot that reads a boring script. They used machine learning to literally dissect the audio of historical scam calls.

SPEAKER_2 (4:00)

And translating human intuition into a machine learning problem. I mean, that requires mapping the actual anatomy of deception.

SPEAKER_1 (4:09)

What does that look like in practice though, feeding it audio?

SPEAKER_2 (4:12)

Yeah. So when we talk about them using machine learning here, we mean they fed algorithms thousands of hours of recorded scams. They wanted to identify the statistical patterns of a fraudster's behavior.