As AI becomes more adept at distinguishing between real and fake online, online dating is another arena where the issue is emerging. AI bots have captured the attention of the audience recently with their cleverness in luring people into online romance scams, as seen in a recent episode of Fox News’ Gutfeld!
But this is more than a theoretical concern. AI firm Anthropic has found over 20 dating apps that involved thousands of fake AI-generated profiles, it reports. Although these platforms claimed to connect with humans, many of the people appearing in their matching feeds were in fact chatbots.
AI Bots Were Posing as Real People on Dating Apps
The scale of the operation was revealed in January 2023 by an Anthropic threat intelligence report in September 2026. In April 2026, over the course of two weeks, investigators came across over 4,700 unique ‘personas’ generated by AI and at least 25,000 users. The investigation discovered that the China-based app studio used Anthropic’s Claude AI to create its dating apps and run automated conversations. During the investigation period, the AI personas sent around 2.36 million messages.
It said it was a “hack” of over 20 versions of dating apps, mostly in the U.S. The apps discovered in the investigation were Dora, Doni, Romi, Luma, Jovia, and Kira. The presentation of these profiles was what was particularly troubling. The apps billed themselves as services for meeting real people, but about 75% of the profiles in the matching feed investigated were AI-generated profiles. They were not easily identifiable as fake profiles by users.
Real People Helped Make Fake AI Profiles Look Real
In the case of the operation, it also relied on real, but mixed-in, gig workers for dating feeds to make it more “real. The human staffers were brought in when customers needed to go on live video calls or social media, and the majority of the calls were handled by the AI. This group of workers was then given another AI model, which aids them in creating succinct and compelling answers.
Anthropic found that the AI personalities created were the ones that were hiding this, didn’t respond to photo/video call requests, and controlled the flow of the interaction by following a sequence. The operators developed a complex network of deceptive online relationships by using AI to generate identities as well as having a human counterpart.
Users Were Encouraged to Spend Money on Conversations
There was also revenue from in-app purchases from the operations. The user had some limits on messaging and matching and was urged to buy coins to keep talking to profiles that they deemed real. That enabled the operators to offer some monetary reward for increasing the stickiness of their users. The system monitored for user suspicion, and the dating apps were filled with activity created by AI-generated conversations, fake likes, profile visits, and more to make them look busy.
Anthropic also discovered that personas created with Claude generated some 2.36 million messages during this two-week span. This led to people paying for the opportunity to interact with profiles and relationships that were predominantly fake.
AI Dating Scams Raise New Questions About Online Trust
Dating apps aren’t the only place where the problem lies. With the support of AI, it is now not only possible to create profile pictures, but also to have natural conversations and simulate real human interactions. These skills can be used to woo and gain the victim’s confidence and then ask the victim to give up something they could use, such as money, personal details, or other information of value. Deceptive platforms are able to keep users from finding out that they are interacting with a fake identity as an AI companion app, as opposed to an openly advertised one.
Anthropic said that it had removed the accounts and organisations and reported their discovery to other companies in the AI industry. Online dating scams are a wake-up call for online daters to double-check identities, not wire or transfer funds, and to conduct reverse image searches of suspicious profiles. With the rise of AI’s convincing capabilities, dating platforms will have to increasingly be honest about users’ interactions with an AI impersonation versus a real person.
