Apate has decided to fight online scams with their own weapons. Instead of just trying to block suspicious calls, the company puts AI agents in the role of victims. These bots answer the scammers, chat them up, pretend to believe the story and keep them busy for as long as possible. Meanwhile, they gather information on how digital fraud works in practice. It is a very concrete use of AI against scammers, one that turns the criminal pitch itself into a source of intelligence.
Apate is an Australian company tied to cybersecurity research at Macquarie University in Sydney. The logic behind it is simple. Scammers rely on volume and time. Every minute spent talking to a convincing bot is a minute not spent targeting a real person. And every call can reveal a little more of the script used to fool whoever is on the other end of the line.
It sounds like something out of a movie. But the idea is pretty down to earth.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
How the AI agents fool the scammers
The trick is sounding human. The agents play victim profiles, with a voice, a way of speaking and reactions that feel real. They hesitate, ask questions, show doubt and, at the right moment, seem ready to follow the instructions of the criminal. To the person running the scam, everything points to a fish on the hook.
But every conversation becomes study material. The goal is to record the script being used, the pressure tactics, the links sent, the payment details requested and the patterns that repeat from one scam to the next. Over time, all this information tends to build a detailed picture of how these criminal groups operate, something that would rarely show up just by blocking phone numbers.
This kind of intelligence can be valuable for those on the front lines, like banks, phone carriers and payment companies. Spotting early which account is receiving money from victims, or which link just started spreading, helps them act before the fraud takes off. The idea ties into another trend in the industry, training and testing agents in controlled environments, as we covered in the case of Patronus AI and its digital worlds for AI agents.
What this means for people in Brazil
In Brazil, phone scams, fake WhatsApp messages and tampered payment slips are part of daily life for a lot of people. Anyone who runs a business or works in marketing and tech knows the damage fraud does to customer trust. One case involving the brand name is enough to flood customer support and give the reputation a real hit.
The Apate case shows that AI agents are not just for customer service or sales. They can also play defense. There is no sign the service works in Brazil, but the idea points to a trend that matters to any company holding customer data. It makes sense to follow this alongside the debate on agents, access and secure identity, a topic likely to grow as more companies put AI in direct conversation with the public.
A lesson that goes beyond security
Even if you do not work in cybersecurity, there is something to take from this. Apate took a familiar problem and changed the angle: instead of only protecting itself from scammers, it started studying them. Picture an online store that takes a close look at the fake messages using its name. With that mindset, it gets easier to warn customers about the latest scam, adjust official messaging and make it clear which channels are legit. No fancy bot required, just an eye for patterns.
The limits that still show up
There are still open questions. It is not clear how well these agents hold up in long, unpredictable conversations, or how the collected data is handled from a privacy standpoint. In Brazil, any similar project would need to pay close attention to data protection laws.
Another question is whether scammers will start using AI to tell when they are talking to a bot. If that happens, a kind of arms race could break out, with each side trying to stay one step ahead. These are common issues when technology leaves the lab and faces the real world, something we have already discussed when looking at the maturity of agents in business processes.
For now, Apate points to a creative way to use AI against scammers: make the criminal waste time and hand over clues without noticing. Whether the idea will scale, and in which countries, is still too early to say. Método Viral will keep an eye on this topic.
