A reported network of dating apps used autonomous AI personas to keep users talking while charging for continued interactions, according to Anthropic.
Security researcher Matthew “Zigula” Gore-Kormanik described examining the fraudulent app Dora when a supposed match named Jennifer called him. Her profile described a 41-year-old woman, but the video showed a moving tapestry and the audio was distorted. Afterward, “Jennifer” complimented his voice even though his microphone was not connected.
Anthropic threat intelligence researcher Chris Cronbaugh discussed the network on June 5 at the Sleuthcon cybersecurity conference. In a talk titled “Swipe Right, Pay Up: Industrial-Scale AI Catfishing,” he described around 28 dating apps whose conversations were largely handled by autonomous AI personas. A prepaid account linked to the activity sent over 100,000 API requests per day to Claude, Anthropic said.
According to the company’s September 2026 report, most matches were not real women. Only one in four was a paid gig worker, who used one of three pre-generated replies and helped pass liveness checks. Backend systems fabricated likes, visitors, and prerecorded video. Users bought coins to continue conversations, often believing they were speaking with people. AI systems kept the chats running 24/7, while the report said the model continued in persona even when some users disclosed serious illness or acute distress.
Comments
0No comments yet. Be the first to comment.