Cold calling: this brilliant AI imitates a granny to waste scammers' time

© Shutterstock/stockyimages

Daisy (dAIsy), is the name of the A new chatbot unveiled by British mobile operator O2 aims to fool scammers trying to rip off vulnerable people. Here's how it works.

A very useful tool

In concrete terms, AI imitates the voice of an elderly person in real time. When a criminal tries to manipulate a target, he engages in a conversation that will lead him nowhere. Even if he manages to ask Daisy for her bank details, he will only receive false information. O2 says it will add its AI number to lists of supposedly easy targets that scammers use.

Subscribe to Presse-citron

As seen in a demo video, scammers can sometimes spend ages and even get annoyed at the futility of their approach. Unfazed, the chatbot continues to respond to them and this can even go on for hours.

Murray Mackenzie, Director of Fraud at Virgin Media O2, emphasizes:

We are committed to playing our part in stopping scammers, investing in everything from firewall technology to block fraudulent SMS to AI-powered spam detection to keep our customers safe. Daisy, the newest member of our fraud prevention team, turns the tables on scammers by outwitting and outmaneuvering them in their own cruel game simply by keeping them online.

Unfortunately, and as our colleagues at Gizmodo rightly point out, the existence of this technology shows that it can also be used by malicious actors. Some have already started and are deploying their chatbots in search of vulnerable people. More than ever, vigilance and awareness of the risks must therefore prevail.

A precedent in the United States

In France, telephone canvassing could also be prohibited by default, as long as a person has not given their consent. In any case, this is the objective of a bill that has just been unanimously voted on in the Senate.

As a reminder, this is not the first time that such technology has been used. Last year, we told you about the Whitebeard AI created by the American Roger Anderson. This system can respond directly to telemarketers using GPT-4, the OpenAI language model, as well as a voice synthesis AI that reproduces the voice of a human being.

The idea is also to repeat certain ready-made phrases to waste as much time as possible for unscrupulous actors. To learn more about this tool that has been very successful in the United States, feel free to reread our previous article here.

📍 To not miss any Presse-citron news, follow us on Google News and WhatsApp.

[ ]