Artificial intelligence now shows up regularly in language classrooms. People usually see it as just technical help, not as any social actor. Studies have given little notice to the way learners treat AI conversation partners as gendered figures or carry their usual cultural habits into those exchanges, especially inside EFL settings. The present work took that gap as its starting point. The study looked at Iranian upper-intermediate EFL learners while they held text chats with chatbots that carried clear male or female identities. Focus stayed on changes in word selection, stance toward the other speaker, and how learners placed themselves in the talk. Performance scores were never the point. The aim was to watch what happened to language once gender signals appeared in a non-human partner. Thirty learners joined, ages 18 to 35. Fifteen were women; fifteen were men. All shared upper-intermediate proficiency. Each person did two fixed chat sessions: one with the male AI, one with the female AI. Thirty full transcripts resulted. Tasks kept a set structure designed to push normal back-and-forth. The analysis leaned heavily on qualitative methods. Feminist Critical Discourse Analysis guided the work, along with Identity Theory and Intercultural Communication Theory. Frequency counts of selected features appeared only to back up the main observations. Gender registered fast. The female AI drew politer forms, cautious phrasing, and friendly gestures. The male AI prompted blunt statements, less softening, and heavy task orientation. Iranian cultural patterns held firm. Indirect expressions, respect for status differences, and face-saving moves moved straight into the AI conversations.AI language practice does not wipe out gender or culture. Learners bring those frames along and keep using them. EFL programs that add AI without close inspection stand to lock in old gender expectations within supposedly modern tools. Close examination of these patterns remains necessary.