Generative AI has undeniably ushered in an era of unparalleled efficiency and convenience. Yet, even the brightest innovations come with their share of inconvenient truths. For Generative AI, these come in the form of cunningly engineered AI-generated scams.
As Generative AI evolved from generating mere text to hyper-realistic images, videos, and voices, so too did the sophistication of scams. According to the Global Anti-Scam Alliance, online scams siphoned an astonishing $1 trillion from unsuspecting victims worldwide last year. In the United States alone, 2023 saw a staggering $12.5 billion in losses reported to the FBI—an unenviable record.
Closer to home in Southeast Asia, the statistics are equally grim. A recent United Nations report highlighted that up to $37 billion vanished into the abyss of romance investment schemes, cryptocurrency fraud, money laundering, and illegal gambling. The same report also pointed an accusing finger at Generative AI and deepfakes, noting their significant role in elevating scam success rates to chilling heights.
Governments across the globe have tried their hand at curbing these digital misdeeds through regulations and punitive measures. Yet, while legislation inches forward, a leading telecom provider in the UK has taken a decidedly more immediate—and intriguing—approach to tackling this modern menace.
UK Telecom Provider Uses Generative AI Agent to Counter Scammers
Virgin Media O2 has unleashed "Daisy," an AI-powered scammer’s worst nightmare. Daisy poses as a chatty elderly woman—scammers’ favorite demographic—and keeps them on the line with endless tales of family drama and knitting patterns. The goal? Waste their time so they can’t waste yours.
Daisy’s brilliance lies in its execution. It transcribes the scammer’s speech, generates suitably rambling responses using a custom language model, and converts them back into lifelike speech. The result is a seamless, real-time interaction that convinces scammers they’ve found their mark, only to lead them into conversational quicksand.
But how does Daisy lure scammers in? No, O2 isn’t snooping on your calls. Instead, it collaborates with experts to release Daisy-linked phone numbers into scammer databases. Predictably, scammers target these "easy" marks, only to find themselves ensnared in Daisy’s endlessly meandering stories.
While not a universal fix, Daisy is a clever addition to O2’s anti-fraud efforts, which have already blocked over £250 million in fraudulent transactions—one every two minutes. With Daisy in the fight, scammers might finally get a taste of their own time-wasting medicine.
Watch Daisy in action against scammers below:
The Future of Fighting Scams: AI vs. AI
As Generative AI continues to evolve, so too does its dark counterpart: AI-generated scams. However, the same technology that powers these scams holds the key to defeating them. Just as Daisy represents a first step in automated scam-baiting, there’s immense potential for AI to tackle scams in other innovative ways—tools capable of detecting deepfake voices and videos, or apps offering real-time scam detection during communications.
While many of these ideas remain at the theoretical or enterprise level, the rise of consumer-focused solutions like Daisy seems inevitable. Why? For one, the market is a wide-open blue ocean, ripe for innovation. For another, governments are already demonstrating the power of AI in fraud prevention. Take the U.S. Treasury’s Office of Payment Integrity, for example. By integrating machine learning into its processes, it has recovered over $4 billion in fraudulent payments in fiscal year 2024 alone. Such successes highlight AI’s growing potential in the fight against scams.
The future may well see a proliferation of AI-driven defenses, creating an ecosystem where scammers have to contend with technology every bit as cunning as their own. And if Daisy’s success is any indication, the era of AI fighting AI is just beginning.
Comments