AI scambaiting is moving from online jokes to a real security debate as scam losses rise.
A viral X post by Vivek Sen has pushed AI scambaiting back into the spotlight after claiming that a user let an AI agent handle scam texts for a week. The post said a scammer asked for a $500 gift card and that the agent dragged the conversation out with fake errands and delays. The claim appears to trace back to a Medium post by Ajay Kumar, who said he used an off-the-shelf AI setup rather than custom code. The full X post could not be independently verified from X, and the Medium post does not provide message logs, code, or a reproducible setup.
That gap matters. AI scambaiting sounds useful because it flips the usual scam dynamic. Instead of a person wasting time with a fraudster, a chatbot does the talking. In theory, that keeps scammers busy and away from real victims. In practice, it also raises questions about proof, safety, and whether ordinary users should be encouraged to engage with criminals at all.
There are verified examples of companies testing this idea. Virgin Media O2 launched Daisy, an AI “granny,” in November 2024 to answer scam calls and keep fraudsters on the phone. O2 said Daisy combined multiple AI models, responded in real time, and had kept scammers on calls for 40 minutes at a time. The company also told customers to leave scambaiting to AI experts and report suspected scam texts to 7726.
The wider fraud problem is large enough to explain the interest. The FTC said U.S. consumers reported losing $470 million in 2024 to scams that started with text messages, five times the amount reported in 2020. The FBI said its 2025 Internet Crime Report logged more than 1 million complaints and nearly $21 billion in cyber-enabled crime losses. It also said AI-related complaints cost Americans nearly $893 million.
At its core, AI scambaiting uses AI chatbots to waste scammers’ time and highlight how automated defen de rs might work.The positive case for AI scambaiting is simple. If automated systems can safely absorb scammer time, carriers and fraud teams could reduce the volume of live attacks. A related research paper on scam-baiting calls found that published scambaiting conversations can help researchers study scam scripts and social engineering tactics. That kind of analysis may improve detection tools.
The risk is that the same technology works in both directions. A 2025 paper titled “ScamAgents” found that autonomous AI agents could generate realistic multi-turn scam calls, maintain memory, adapt to user responses, and convert scripts into lifelike voice calls. The author, Sanket Badhe, argued that current safety tools are weak against agent-based deception spread across several turns.
The AI Decode has also covered how AI-enabled fraud can exploit trust without needing a technical system breach, especially in deepfake and voice-cloning cases. That makes AI scambaiting a narrow tool, not a full defense. The next question is whether telecoms and security firms can use it without encouraging risky amateur copycats.
Regulators and telecom operators are also watching scambaiting closely. In India, the Reserve Bank has launched digital awareness campaigns to warn consumers about AI-driven scams. Security researchers note that deliberately interacting with scammers can expose personal data if not done carefully. Because scambaiting uses up time and computing resources, carriers like Verizon and BT are studying whether automated bots are more efficient than simple blocking mechanisms. Future tools might integrate with call-blocking apps and messaging platforms, providing context to law enforcement while anonymizing user identities
Ultimately, AI scambaiting is only as effective as the awareness and caution of the people using it. AI scambaiting should complement, not replace, reporting scams and protecting personal information.
For now, the best practice is to avoid responding to unknown messages or calls and to report them to authorities. Readers interested in understanding how deepfakes and voice-cloning scams operate can read our earlier coverage on AI-enabled f9raud and the $25 million deepfake scam on a
For deeper insights and tips on AI scambaiting, follow The AI Decode’s continuing coverage of AI scambaiting technology and cybersecurity trends.
