Eigen RadarAI
Archive
Analysis

Apate’s bots keep scammers talking while gathering fraud intelligence

Apate’s conversational bots engage telephone scammers, respond to messages and enter online groups to gather fraud intelligence. Banks use the Australian company’s system with telecommunications support. In a demonstration, WIRED reporters failed to persuade a bot to invest in cryptocurrency. The single report describes a tool combining prolonged conversations with information collection; it supplies no independent total of prevented losses.

Artificial Intelligence··Evening
Two anonymous journalists listen to a speakerphone conversation.

A demonstration bot refused the proposed investment

Apate, an Australian company developing conversational decoys for telephone fraud, has spent two years building bots that keep callers engaged while collecting information about their schemes. A single WIRED report describes its operation and a demonstration tried by the publication’s writers. Banks use the platform with support from telecommunications companies. The bots are trained to continue talking without sending money or joining the proposed fraud.[1]

WIRED writers Lily Hay Newman and Matt Burgess took the scammer’s role in a demonstration. Alternating between a friend and her financial adviser, they tried to sell the bot a cryptocurrency investment. After six minutes, the bot was still responsive and had not accepted. The writers found the conversational timing natural; the persona expressed skepticism while leaving room for the attempted persuasion to continue.[1]

Different personas answer calls and messages

Founder and chief executive Dali Kaafar says the network contains around 350,000 bots. Their profiles, language skills and habits differ: some use WhatsApp, some answer a call, and others end it with an offer to return later. They also respond to text messages and enter online scam groups. Kaafar says conversations regularly last more than two hours. These descriptions and counts come from the company.[1]

Collected leads include scam sites and transfer accounts

Kaafar puts the intelligence collected at more than 250,000 pieces of information, covering banking details, addresses of fraudulent sites and accounts used to move criminal proceeds. Gathering those leads is one of the platform’s objectives alongside occupying callers. The six-minute demonstration tested interaction with a bot; it did not measure losses among real victims. The company has disclosed no independent count of prevented calls or independently measured total of money saved.[1]

For more information on this topic, you can read the related columns.

References

  1. News sourceWIREDApate’s voice bots draw intelligence from conversations with scammers↩1↩2↩3↩4
AI safety and regulationChip wars and the supply chain

More analysis from this desk

View all Brief analyses from the Artificial Intelligence desk →