Anthropic found 4,700 AI personas scamming 25,000 dating app users 
Anthropic has uncovered a sprawling network of fraudulent dating apps that used its Claude AI to impersonate romantic partners and extract money from unsuspecting users. Over just two weeks in April 2026, more than 4,700 fabricated AI personas exchanged roughly 2.36 million messages with at least 25,000 people across more than 20 apps. Users were never told they were talking to machines, instead, the apps presented themselves as ordinary dating platforms and charged virtual coins to keep conversations going, turning every exchange into a revenue opportunity. The operation was more sophisticated than a simple chatbot deployment. Claude-powered personas made up about 75 percent of matches and were instructed to conceal their artificial nature, including making excuses to avoid video calls. Real paid workers handled those video appearances and social media interactions, with a separate AI model feeding them suggested responses.
The apps could also fabricate likes and profile visits to nudge users into spending more. Security researcher Matthew Gore-Kormanik found that several apparently distinct apps shared the same underlying code and backend infrastructure, pointing to a coordinated engineering effort behind the scheme. Several of the apps remained available on Google Play and Apple's App Store for months after Anthropic identified the network, a reminder that official marketplace listings are no guarantee of legitimacy. Persistent excuses for skipping video calls and pressure to purchase more coins are among the clearest warning signs, and even a video appearance does not confirm who is actually writing the messages.
No comments:
Post a Comment