According to Japan’s Asahi Shimbun, British news agency Virgin Media O2 has developed and is operating an AI chatbot called Daisy to block voice phishing crimes.
Daisy is a voice-based AI set as a 78-year-old elderly living alone, and the latest large-scale language model (LLM) and speech synthesis technology are applied.
Daisy listens to what the other person says in real time and naturally continues the conversation. The purpose is simple. She wants to make sure she talks to a con artist for as long as possible to waste her time.
O2 secretly included Daisy’s phone number in the list of fraudulent people distributed illegally by voice phishing organizations. Criminals who do not know her make phone calls because they judge her as a gullible old man, and Daisy delays calls by pretending she didn’t understand or changing the subject with absurd stories.
Cases of actual phone calls have also been made public. When a scammer demanded that the app be installed, Daisy talked about bread or knitting, and when the other person got annoyed, she responded, “When you get older, you talk more.”
Daisy has spoken to more than 1000 crooks in about a year since its November 2024 release, according to O2. Some of the calls lasted more than 50 minutes. During this time, the likelihood of a real victim being called during a con man’s call with Daisy was reduced by that much.
This method is called “scambaiting” technique, which deceives con artists in reverse. O2 explained that AI cannot completely prevent crime, but it is meaningful in lowering the possibility of damage.
“The most effective response is to immediately cut off suspicious calls and report them,” O2 said, reiterating personal alarm. The project won the Best AI Utilization Award at the British Marketing Awards last year in recognition of its achievements.
JENNIFER KIM
US ASIA JOURNAL
