A bot on the dating app Tinder directed men to flirt with one another, in the belief they were communicating with a woman.
According to reports, a developer was able to manipulate Tinder’s application program interface (API) to create a bot that appeared to be a woman, but which would relay messages between men who had liked ‘her’ profile picture and initiated a conversation with it. “Tinder makes it surprisingly easy to bot their system,” the programmer, described as ‘Patrick’, told The Verge. “As long as you have a Facebook authentication token, you can behave as a robot as if you were a person.” He monitored the ensuing conversations and stepped in if it looked like the men were arranging to meet in real life.
Like many popular apps, Tinder has seen its private, internal API replicated and used by third parties to create bots. In many cases, these bots initiate automated chats (using generic, pre-typed responses) to lure users into engaging with paid products and services. Security firms such as Symantec have documented such issues for several years. One developer, Justin Long, built an application that constructed facial models based on those photos a user had previously ‘liked’ or ‘disliked’, and then automatically initiated conversations with those that fit the facial profile.
The bots pose a business risk to Tinder. If human users find themselves wasting time talking to bots, they are likely to lose interest in the app. In order to prevent either bots or humans from automatically ‘liking’ every image they come across, on 12 March the company announced an algorithm that limited the number of ‘likes’ a user can apply within a 12 hour period. In a blog post, Tinder said this simple measure had led to what it called “a 52% decrease in spam bots”, which it said were one of its biggest user complaints. However, users who paid for Tinder Plus would have a higher cap on ‘likes’, it said.