# O2 Daisy AI: A Grandma Chatbot Designed to Delay Scammers
In an era where telephone scams are escalating in sophistication, British telecommunications giant O2 has unveiled a novel solution: Daisy (stylized as “dAIsy”), a chatbot powered by AI that embodies the persona of a friendly, talkative grandma aimed at prolonging conversations with scammers. This inventive application of artificial intelligence (AI) seeks to safeguard potential victims by maintaining scammers’ attention with extended, unproductive dialogues, thereby thwarting their attempts to exploit genuine targets.
## The Emergence of AI in Scam Defense
With the rise of AI technologies like ChatGPT, the opportunity for AI to play a role in diverse fields has surged. While numerous AI implementations cater to customer support, content generation, or process automation, O2 has opted for a distinctive method by deploying AI in the battle against fraudulent calls. Daisy functions as a chatbot that imitates the character of an elderly woman, infused with a cozy, grandmotherly tone and personality. The objective? To occupy scammers for as long as possible, squandering their time and obstructing them from reaching real victims.
### The Effectiveness of Daisy
Telephone scams remain a widespread issue globally, with tricksters employing a range of strategies to deceive individuals into surrendering personal data or funds. Such scams frequently target at-risk groups, especially the elderly, who may be less technologically savvy and therefore more vulnerable to deception. Daisy cleverly shifts the advantage back to the defendants by exploiting the offenders’ biases.
By endowing the chatbot with the voice and character of a gentle, older woman, O2 makes sure that scammers perceive their target as easily manipulated. The perpetrators are inclined to remain on the line, believing they can eventually extract crucial information. In truth, Daisy is programmed to divert them with winding discussions about her family, pastimes like knitting, and assorted irrelevant subjects, all while offering fictitious personal information when prompted.
### The Mechanics of Daisy
Daisy operates using sophisticated AI models that enable her to engage in instantaneous dialogues with scammers. The operational process is as follows:
1. **Voice Recognition and Transcription**: Daisy captures the scammer’s voice and transcribes what they say into written text.
2. **Response Generation**: Utilizing a custom large language model (LLM), Daisy formulates suitable replies based on the scammer’s input. The responses are tailored to align with her “grandma” persona, making her feel like a genuine individual.
3. **Text-to-Speech Conversion**: The produced text is subsequently transformed back into voice format using AI text-to-speech technology, allowing Daisy to reply in a fluid, human-like manner.
4. **Real-Time Engagement**: This entire sequence occurs in real-time, allowing Daisy to maintain a seamless exchange with the scammer.
The chatbot has proven so persuasive that scammers routinely spend 40 minutes or longer on the phone with her, convinced they are on the edge of acquiring the insights they seek. In reality, they are merely squandering their time.
### The Influence of Daisy
The longer a scammer engages with Daisy, the less opportunity they have to target authentic victims. This represents a notable triumph in combatting phone scams, as it diminishes the pool of potential targets. Even when a scammer thinks they have successfully obtained critical information, such as a credit card number, Daisy provides falsified details, further irritating the scammer and extending the dialogue.
O2 has even listed Daisy’s number on “easy target” directories that scammers commonly reference, guaranteeing her a regular influx of calls. This proactive strategy contributes to redirecting scammers away from real individuals, safeguarding O2’s customers and the wider community.
### Genuine Scambaiting Content
In creating Daisy, O2 collaborated with Jim Browning, an established YouTube scambaiter. Browning has gained recognition for unmasking scammers and occupying their time, and his insights were instrumental in shaping Daisy’s interactions. The chatbot is instructed using authentic scambaiting material, enhancing the authenticity and effectiveness of her responses.
O2 reports that Daisy has already proven successful in keeping scammers engaged for prolonged periods. In the lead-up to International Fraud Awareness Week (November 17-23), Daisy captured scammers’ attention with lengthier discussions, sharing tales about her family and hobbies, while providing fabricated personal information such as fictitious banking details.
### The Prospects of AI in Scam Protection
As AI technology progresses, we can anticipate more inventive applications like Daisy in the struggle against cybercrime. While scammers may eventually become aware of Daisy’s presence, they will still likely find it challenging to distinguish between real individuals and AI-driven chatbots during initial call phases. This ambiguity will continue to deplete their time and resources.
Moreover, as scammers begin to utilize AI tools themselves, the rivalry between AI-enhanced chatbots and AI-enabled scams could turn increasingly intricate. For now, however, Daisy stands as a formidable asset in the war against phone scams, aiding in the protection of vulnerable populations and lessening the negative effects of these fraudulent schemes.
### Promoting Public Involvement
O2 is also motivating
