This study examined how teachers’ perceptions of anthropomorphic qualities in educational artificial intelligence shape their views on professional role boundaries. The study used a crosssectional survey design. The population consisted of in-service teachers with prior experience using conversational artificial intelligence. A sample of 412 teachers was selected through a multistage sampling procedure that involved stratification by school type and simple random selection within each stratum. Data were collected with a structured questionnaire that covered perceived anthropomorphism, boundary perceptions, and background characteristics. The instrument was reviewed by three experts for content validity. Internal consistency was confirmed through a pilot test, which produced reliability coefficients ranging from 0.78 to 0.86. Data were gathered in person with the support of trained assistants. The completed forms were screened, coded, and entered into a statistical package for analysis. Descriptive statistics were used to summarise the variables. Hierarchical multiple regression was used to test the predictive strength of perceived anthropomorphism on boundary perceptions, while controlling for gender, years of teaching, frequency of AI use and subject area. The analysis showed a clear and statistically significant link between anthropomorphism scores and boundary perceptions (p < .05). Teachers who rated the systems as more human-like reported greater uncertainty about the limits of their professional roles. AI usage frequency had a small but significant effect, while subject area, years of experience, and gender did not. The study concluded that anthropomorphic cues in educational systems influence how teachers judge task boundaries. It was recommended that teacher training programmes should provide guidance on responsible use of AI. Keywords: Perceived anthropomorphism, Educational artificial intelligence, Teacher professionalism and Perceived professional boundaries