The anecdote in the claim is a near-verbatim restatement of a passage in a real, published article. Full-text republications of the article report that Gore-Kormanik was analyzing Dora when a call from "Jennifer" arrived, that he answered and saw a tapestry that was moving, probably due to a fan, and heard weird distortion in the background, after which "Jennifer" messaged that she'd had fun and "your voice is way better than expected," despite his microphone not having been connected . The same article records that the researcher was examining these apps because the reporter had shared a spreadsheet of suspected scam apps with him.

The surrounding context in the same article is important and the claim omits it. The article reports that in this network, the gig workers were hired to pass liveness checks on video or to follow social media accounts, and they did not write their own messages but selected from three pregenerated replies , while the AI personas could only demur with a plausible explanation for why a video chat or call was not possible . It also quotes Anthropic to the effect that backend components fabricated likes, visitors, and pre-recorded "video" when no real person was available, and tracked which users had begun to suspect they were talking to a bot . The article further reports that Gore-Kormanik recovered an internal repository containing notes for monitoring gig workers to see whether their cameras were on and broadcasting and whether they were reachable by message, and that the app takes screenshots and records calls, which are transcribed .

The broader network is documented by a vendor threat report. Secondary reporting states that Anthropic said in its "Detecting and Countering Misuse of AI: September 2026" report, released September 10, that a Chinese app studio used Claude to build and operate a network of more than 20 dating apps while advertising them as platforms for interacting with real people, and that over a two-week period in April 2026 it identified more than 4,700 distinct AI personas that had conversations with at least 25,000 people . The architecture described is hybrid: roughly 75 percent Claude-powered personas and 25 percent real people in the user-facing feed, with the real workers incorporated to provide authenticity checks the AI could not perform, including live video calls and social media follow-backs . The Verge article reports a larger figure from the conference talk: Cronbaugh described a network of around 28 dating apps where the conversation was largely run by autonomous AI personas, and said the majority of chats did not involve a human agent at all . The talk itself is confirmed on the conference organizer's own site, which describes an investigation into a fraudulent dating app network that abused AI to deceive victims at industrial scale, using real apps, on real app stores .