Australian startup Apate deploys AI agents that pretend to be potential victims so scammers reveal infrastructure that banks, telecom companies and government agencies can block. The agents answer calls, messages and email, follow phishing links and enter online groups using fabricated identities and multiple voices and languages.
Their purpose is not only to waste a scammer’s time. Apate tries to collect bank-account numbers, cryptocurrency wallets, phone numbers, malicious websites and impersonated brands before a real victim loses money. Customers receive the resulting intelligence through an analytics portal and can assign a bot a specific target, such as obtaining a wallet address.
The company says its agents have conducted 2.5 million conversations and occupied scammers for a combined 17 years. It also says TPG Telecom diverted nearly 590,000 calls to the system and Commonwealth Bank of Australia saw a 76% decline in publicly reported scam losses. These figures come from Apate and do not isolate every factor that may have affected fraud rates.
Apate uses frontier and internally fine-tuned models trained partly on prior scam conversations. The 20-person company raised $8.15 million in August and plans to expand in the UK and US. Its approach could surface threats earlier, but fabricated identities and automated engagement still require legal, privacy and operational oversight.