The migration of public communications to the internet and the development of artificial intelligence have created favorable conditions for manipulating public opinion. Currently, the massive presence of social bots in public discussions disrupts the digital cultural-communicative environment, undermines trust in online discussions, and distorts the reflection of real public sentiments. This phenomenon poses a significant challenge to democratic discourse and informed citizenship. Users of social networks, including the Russian platform VKontakte, struggle to identify automated accounts and often engage with them in dialogue as if they were live interlocutors. A contradiction emerges between the rational recognition of a bot's programmatic nature and the performative attribution of subjectivity to it, a tension that remains largely unnoticed by ordinary users, which calls for a specific cultural analysis of speech practices of accusing others of being a "bot". The situation is complicated by the fact that the social network contains accounts of real people who are functionally indistinguishable from those managed by neural networks. The novelty lies in a comprehensive cultural, semiotic, and empirical analysis of a phenomenon that has previously been studied primarily through technical or sociological methods. This study is the first to analyze the speech practices of real users interacting with bots in the social network VKontakte. The results of the study show that users poorly recognize bots, and when they do identify them, they respond emotionally, anthropomorphizing the bots. Unlike in foreign social networks, within VKontakte the term "bot" is almost never used as an insult against real people.