AI agents posing as easy marks: Australia's Apate keeps scammers on the line, across 2.5 million conversations so far

Anyone who has had a call from "the bank" or a message about "a parcel waiting" knows the rule: hang up and block. Australian startup Apate does the opposite. Its AI agents answer the calls, open the suspicious links, join groups on social media and play naive victims - of different ages, genders, accents and languages. The goal is to keep the scammer on the line long enough to reveal the accounts, crypto wallets, phone numbers and fake websites they work with.

The company was founded in 2024, based on research at Macquarie University in Sydney, and has 20 employees. According to its own figures, its agents have so far wasted around 17 years of scammers' time, across 2.5 million conversations, and they hold tens of thousands of new ones every day. Clients can also give them a specific mission - for example, extracting a bank account number or a crypto wallet address.

The idea was born at a family picnic. Founder and CEO Dali Kaafar got a call from a scammer and, to entertain the kids, kept him on the line for 44 minutes. That's when he realised those were 44 minutes in which the scammer wasn't robbing a real person. Co-founder Brad Joffe says the industry reacts late: "Most of the industry only acts after the damage has already been done." The name, he explains, comes from Apate, the ancient Greek goddess of deceit - "fitting, given what our agents do".

Clients include the Commonwealth Bank of Australia, the operator TPG Telecom, utilities and government agencies. TPG has redirected almost 590,000 calls to Apate's agents. Commonwealth reports 76 percent fewer publicly reported scam losses, and an unnamed Australian bank used the agents to disrupt 22 scam campaigns. In August the company raised 8.15 million dollars (around 7 million euros) in seed funding, just under 10 million in total, and it is now opening an office in London and expanding in the US.

Why this is more than a fun story: the scammers use artificial intelligence too. In 2025 the FBI received more than 22,000 reports of AI-related crime, with losses of around 893 million dollars - more than 632 million of that from investment fraud alone, using fake AI videos of celebrities and executives. A race is opening up in which both sides deploy lying machines. Joffe says it without a hint of embarrassment: "We're probably one of the few businesses in the world that celebrates when people swear at our product."

The figures come from the company itself and are hard to verify from the outside. But the model is clear: instead of every citizen defending themselves alone, someone systematically steals the scammers' time and gathers evidence against them. Would any bank or telecom operator in the Balkans invest in a defence like this - or will we keep relying on "hang up and block"?