Developed by UK provider Virgin Media O2 and announced on Thursday, "Daisy" is an AI-fuelled call answering service that aims to keep scam callers on the line as long as possible, meaning less time spent with potential human victims. It's the same idea we've seen in a fair few time-wasting bots in the past — and it's the signature strategy adopted by scam fighter Scamalot aka James Veitch.
O2 worked on the AI with YouTuber Jim Browning, whose scambusting work has seen him track and expose many a fraudulent scheme using various strategies.
Described by the company as "head of scammer relations", the Daisy AI is programmed to give rambling stories to callers — and I'm not going to lie, the details sound a little bit like age-based stereotyping of elderly women but who am I to say what a scammer will believe? According to O2, Daisy has told "meandering stories of her family, talked at length about her passion for knitting and provided exasperated callers with false personal information including made-up bank details." The company claims Daisy "has successfully kept numerous fraudsters on calls for 40 minutes at a time."
According to O2, Daisy is the result of multiple AI models that listen to the caller and make a live transcription. Then, the program generates an appropriate response from its language model, delivered in a human-like voice embedded with Daisy's personality.
You won't be able to interact with Daisy yourself (unless you're a scammer). When I reached out to the company for further information, an O2 spokesperson told me, "The purpose of creating Daisy was to both waste scammers time and to create a campaign to educate the public on the danger of scam calls. The tool was purpose built to interact with scammers and so is optimised to do that rather than have general conversations. Opening the tool up to everyone would also require a huge amount of computing power, so right now this isn’t something Daisy is able to do."
In the case that the scammer makes it through to you instead of Daisy, you can forward suspected scam calls and text messages to O2's existing blocking service at 7726.
Finally, a use for generative AI that I can get behind!
Just two weeks ago, a deep fake attack involving a CEO's voice trying to get his employees credentials happened.
https://techcrunch.com/2024/10/28/wiz-ceo-says-company-was-targeted-with-deepfake-attack-that-used-his-voice/?guccounter=1