<\/p>\n
A Delhi-based IT professional employed generative AI to turn the tables on an army impersonation scam, highlighting a rising trend of tech-savvy scams and DIY cybersecurity countermeasures in India.<\/p>\n<\/div>\n
When a message from a supposed college contact claiming to be an Indian Administrative Service officer arrived on his phone, a Delhi-based IT professional says he decided not to ignore a familiar “army transfer” fraud but to turn the tables using generative AI. According to the original report, the user , posting as u\/RailfanHS on Reddit , asked ChatGPT to generate a webpage that mimicked a payment portal and secretly captured the visitor\u2019s GPS coordinates, IP address and a front\u2011camera photo. [1]<\/a><\/sup><\/p>\n The Reddit account\u2019s detailed thread, which went viral in India, describes how the scammer sent photos of goods and a QR code and demanded an upfront payment. Feigning difficulty scanning the code, the poster sent the scammer a link to a hastily produced PHP page; when the fraudster clicked to “upload” the image the browser prompted for camera and location access and the poster received the live data. “Driven by greed, haste, and completely trusting the appearance of a transaction portal, he clicked the link,” the poster wrote. The result, he says, was immediate panic and messages pleading for mercy. [1]<\/a><\/sup><\/p>\n While the account cannot be independently verified, the technical method was scrutinised and replicated by other Reddit users, who confirmed ChatGPT can produce functional code that requests geolocation and camera permissions as part of a seemingly legitimate upload flow. One commenter said they were “able to make a sort of a dummy HTML webpage” that captured geolocation after asking for permission. The original poster acknowledged using specific prompts to bypass some safety guardrails and hosting the script on a virtual private server. [1]<\/a><\/sup><\/p>\n Experts and broader reporting underscore that this online anecdote sits inside a larger pattern: organised fraud that impersonates army personnel, public officials or law enforcement is widespread in India and beyond. Government and press accounts show multiple recent cases where victims were coaxed into transfers or large payments , from recruitment scams that collected lacs of rupees in Maharashtra to complex confidence frauds involving long\u2011running deception and large sums. Such schemes commonly use social engineering to override victims\u2019 caution. [2]<\/a><\/sup>[5]<\/a><\/sup>[6]<\/a><\/sup>[4]<\/a><\/sup><\/p>\n At the same time, investigative reporting has documented how criminal networks across Asia are weaponising AI at scale. A Reuters investigation found that scammers in Southeast Asia have used ChatGPT to craft tailored scripts for large\u2011scale online fraud operations, sometimes with grave human costs for coerced workers at scam compounds. That reporting highlights how accessible generative tools can be repurposed to increase the reach and persuasiveness of scams. [3]<\/a><\/sup><\/p>\n The Reddit episode illustrates a new, DIY dimension to scambaiting , where tech\u2011savvy individuals deploy the same tools criticised for enabling crime to expose or embarrass fraudsters. Platforms and cyber\u2011security professionals warn that such “hack\u2011backs” occupy a legal grey area and may carry risks, even where motives are to frustrate criminality; independent authorities and courts, rather than private retaliation, remain the recommended route for managing fraud. Industry data and expert comment stress prevention: public awareness, stricter platform practices around permission prompts and rapid law\u2011enforcement cooperation. [1]<\/a><\/sup>[3]<\/a><\/sup><\/p>\n As generative AI continues to lower the barrier to building convincing web pages and social\u2011engineering scripts, the line between empowerment and misuse tightens. According to the original report, the Reddit thread prompted others to replicate the approach, underscoring a broader shift in how both offenders and those who oppose them use AI. Policymakers, platform operators and users will need clearer rules and technical protections to prevent permission\u2011prompt deception while preserving legitimate developer and accessibility uses. [1]<\/a><\/sup>[3]<\/a><\/sup>[4]<\/a><\/sup><\/p>\n ##Reference Map:<\/p>\n\n