This paper develops an interdisciplinary perspective, which combines ideas from anthropology, sociolinguistics and interface design, on how AI chatbots project recognizable social identities. Specifically, it brings together Silvio’s notion of animation, the social practices through which “humanness” is projected onto nonhuman entities, and Blommaert’s notion of enoughness, the idea that the authenticity of linguistic performances is a matter not of the accuracy of a performance but of how audiences collectively evaluate it as socially recognizable. The analysis draws on a corpus of metapragmatic artifacts posted on social media sites related to ChatGPT’s advanced voice mode and Sesame.ai’s hyper-realistic voice interface. Analysis of these artifacts reveals how designers, users and AI systems co-produce boundaries of authenticity through the deployment and uptake of linguistic and discursive features such as accent and stance. In doing so, they continually recalibrate what counts as culturally competent performances, shaping emergent norms of identity and sociality around AI. The paper highlights how humanness and culturality are distributed across technical systems, corporate discourse, and human interlocutors, with important implications for understanding how generative AI reproduces cultural stereotypes by drawing on the linguistic labor of users.