I present a sociologically informed case against human-robot friendships. Some philosophers have attempted to use the sociology of everyday life to draw the opposite conclusion. They suggest that our friends are characters according to the sociological definition of this term (character). When someone plays this character, they regularly conceal their mental states from us to appears friendly. Therefore, we can relate to robots as friends if they competently play the character of a friend. I argue that this position misinterprets the social theory its uses. I examine the sociological literature on passing to forward this claim. People who practice passing play characters that do not accurately represent who they are. Crucially, they stop performing as characters to enter friendships. I argue that this shows that friendships center around the absence of character performances. Therefore, we should reject the idea that we can relate to robots that play the character of a friend a friend – as a friend is not character anyone or anything can play. Additionally, I identify two privacy issues the abovementioned way of interpreting supposed human-robot friendships excuses. And explain how we could accurately use character-focused social theory to describe why some people mistakenly believe robots are their friends.