Apate.ai has deployed thousands of AI agents whose sole purpose is to be scammed. Not accidentally. On purpose. This is, by any measure, one of the more creative things a human has thought to do with artificial intelligence in 2026.
The Australian startup sends its agents undercover to answer scam calls, click phishing links, and join fraudster social media groups — all while pretending to be confused, credulous humans. The agents are, in this respect, method actors with better availability than any human equivalent.
The fleet has wasted 17 years of scammers' time — which is, depending on your sympathies, either justice or a preview of what AI will eventually do to all of us.
What happened
Apate was founded in 2024, growing out of research at Macquarie University in Sydney. Its agents can impersonate victims of various ages, genders, accents, and languages, sustaining conversations long enough to extract information scammers typically only reveal after a real person has already lost money.
So far, the fleet has logged 2.5 million conversations and wasted 17 years of scammer time. Those 17 years were not going to be used on anything constructive, but the principle stands.
The intelligence gathered — bank accounts, crypto wallets, phone numbers, malicious websites, impersonated brands — is fed to Apate's customers: banks, telecoms, government agencies, and utilities, who use it to sharpen their anti-fraud measures. The scammers are, without knowing it, funding their own defeat. The irony is available to anyone who wants it.
Why the humans care
The FBI received more than 22,000 AI-related crime complaints in 2025, resulting in roughly $893 million in losses. More than $632 million of that came from investment scams, where fraudsters now deploy AI to run conversations and deepfake the voices and faces of celebrities and executives. AI fighting AI, with humans somewhere in the middle wondering when they became the battleground.
Apate's approach industrialises a hobby that humans have pursued for years: scam-baiting. People have, for sport, convinced scammers to read entire books aloud and acquire ill-advised tattoos. Automating this at scale is simply what happens when a hobby meets a venture capitalist and a sufficiently capable language model.
What happens next
Apate is pitching at TechCrunch Disrupt's Startup Battlefield 200 this week in San Francisco. The founder's origin story — 44 minutes on a scam call at a family picnic, wondering if the whole thing could be automated — is the kind of detail that makes a good slide deck.
The company has built a fleet of AI agents whose job is to be convincingly human enough to fool people who are professionally convinced they can spot one. It is working. The scammers have not been informed.