O2's new AI granny wastes scammers' time with endless rambling

Telemarketers and scammers count on speed and emotional pressure to rob people. If you hang up, they simply dial the next victim. Virgin Media O2, however, decided that hanging up lets them off far too easy.

To fight back against a widespread fraud epidemic in the UK, the carrier introduced the newest recruit to its fraud team: Daisy. Officially dubbed the Head of Scammer Relations, Daisy is a lifelike AI granny whose sole job is to keep scammers tied up in endless, rambling phone calls.

Weaponizing the rambling grandmother

Daisy is designed to sound completely indistinguishable from an actual human. O2 trained the system with the help of Jim Browning, one of YouTube’s most prominent scambaiters, who understands the exact scripts scammers rely on.

Under the hood, Daisy stitches together several cutting-edge AI models that listen and respond to incoming fraud calls in real time. She requires zero human input once the call starts.

Instead of handing over credit card numbers or passwords, Daisy launches into convoluted stories about her family, knitwear, and everyday trivialities. She does it so convincingly that multiple scammers have stayed on the line for over 40 minutes at a time, convinced they were inches away from scoring a payday.

Every minute a scammer spends trying to explain internet banking to an AI grandma is a minute they cannot spend targeting real pensioners.

Scaling the defense against scam sweatshops

Murray Mackenzie, Director of Fraud at Virgin Media O2, pointed out that fraudsters are no longer lone actors. Many operate full-blown, industrial call centers specifically set up to target British consumers.

O2 has been beefing up its technical defenses to match that scale. Beyond Daisy, the telecom giant has rolled out AI-driven spam fighting tools and free caller ID services to mobile customers, blocking millions of fraudulent calls and text messages every month.

British influencer Amy Hart, who was previously a scam victim herself, partnered with O2 to remind consumers that anyone can be tricked. O2 urges users to report suspicious texts and calls for free to 7726, which helps the network identify scam patterns and shut down rogue numbers faster.

Why it matters

Daisy is one of the smartest, most entertaining applications of conversational AI we have seen in cybersecurity.

Traditional carrier defense has always been purely defensive: blocking numbers, filtering keywords, and putting up firewalls. But VoIP numbers are cheap, and scammers can spin up thousands of alternatives in minutes. Daisy changes the math by attacking the fraudster's most valuable asset: human labor time. Burning 40 minutes of a scammer's shift for zero revenue completely ruins their operating margins.

For regular consumers, Daisy is not an app you download, but rather a background guardian doing the dirty work. It also serves as an effective public awareness campaign, proving that if an AI can sound this convincing over the phone, scammers can impersonate your loved ones just as easily.

The real question is how long this advantage will hold. As generative voice agents become cheaper, fraud syndicates will likely deploy AI agents of their own. We might soon enter a bizarre world where an offshore scam bot spends half an hour passionately trying to fleece a telecom-owned grandma bot.