I was just sipping iced coffee at a roadside tea stall when I came across a report from Anthropic, numbered GTG-15001. The details were mind-blowing. There’s an app studio in China that, in just two weeks, managed to create 4,700 AI girlfriend accounts while simultaneously “crashing” 25,000 profiles of older men overseas. You might think you’re diving into a pool of potential love, but it turns out that 75% of the responses are from AI, with only 25% being real people doing part-time work.
The operation is slick: AI handles all the sweet talk and emotional support, and only when it’s time for a video call—when the AI might slip up—do they let a real person step in. Even when real people reply, the backend generates several message options for them to choose from. This group has launched over 20 dating apps, claiming to offer real matches, but it’s all running on the Claude model for automation, racking up about 2.36 million chat messages. If the older men want to keep chatting, they have to pay to buy credits, essentially wrapping a scam in an AI disguise.
They’ve even gone to great lengths to avoid scrutiny, creating a UI controller that only activates a compliant interface when Apple and Google are reviewing the apps. The rest of the time, it’s completely dormant, and they keep changing the names of the apps to dodge detection. The report also mentioned that the AI company can see all conversations between users and AI, allowing them to investigate, ban accounts, and generate reports without needing approval from regulatory bodies or user consent. Meanwhile, the exhaust from night-shift motorbikes still lingers in the air, while screens in rented rooms are likely filled with the retirement savings of countless older men, slowly being siphoned off by these cold lines of code.
