Cybersecurity startup Apate has begun deploying artificial intelligence agents to pose as scam targets, deliberately engaging with fraudsters to harvest intelligence on their tactics and infrastructure, the company announced.
The approach represents a shift in how fraud intelligence is gathered. Instead of relying on human scam-baiters who volunteer their time to waste fraudsters' hours, Apate's AI agents can operate at scale, handling thousands of simultaneous conversations while logging every technique, phone number, and payment method the scammers use.
The timing reflects an escalating problem. The Federal Bureau of Investigation has reported that Americans lost more than $12.5 billion to internet scams in 2023, a 22% increase over the prior year. AI tools have made it cheaper for criminals to generate convincing phishing messages, clone voices, and run romance or investment frauds that once required teams of operators.
Apate's system works by having its AI agents respond to inbound scam messages or calls as if they were ordinary, vulnerable targets. The agents string scammers along, extracting details about how the operation works — which payment apps they demand, what scripts they follow, and where they route stolen funds. That data is then compiled into threat intelligence that can be used by banks, telecom carriers, and law enforcement.
The concept is a natural extension of "scam baiting," a practice where volunteers waste scammers' time by pretending to fall for their tricks. But human baiters can only handle one conversation at a time and often burn out. AI agents don't sleep, don't get frustrated, and can be deployed across dozens of languages and platforms simultaneously.
Scammers themselves have already adopted AI. Voice-cloning tools let criminals impersonate a victim's relative in a phone call. Large language models generate flawless phishing emails at no cost. That asymmetry — cheap AI for criminals, expensive human defenders — is what Apate is trying to correct.
The company has not disclosed how many agents it operates or which specific fraud networks it has infiltrated. But the model raises questions about consent and entrapment that legal analysts will likely watch closely as the technology spreads.
The global anti-fraud market is projected to reach $63 billion by 2028, according to industry forecasts, as banks and governments scramble to counter increasingly automated attacks. Apate's bet is that fighting AI with AI — and turning the scammers' own tools against them — will prove more effective than traditional defenses.
