Making Ethical AI Teachable: Distributed Accountability in Language Teacher Education

Discussions of ethical generative artificial intelligence (GenAI) use in language education are often disconnected: whether by focusing on practical concerns (classroom dos and don'ts around disclosure, academic integrity, the boundaries of acceptable assistance) or through broader critical questions concerning surveillance, platform power, data extraction, linguistic bias and unequal participation in AI development. This split has practical consequences for language teacher education because it associates ethics with teacher compliance while leaving GenAI's structural conditions unexamined. This viewpoint article proposes distributed accountability as a conceptual and teachable framework for ethical AI praxis. It provides a language‐education‐specific framework that highlights how responsibility is shared by learners, teachers, institutions, AI developers and vendors. The distributed accountability model supports the implementation of three practice‐informed heuristics: a stakes‑and‐harm filter, an authorship receipt and a linguistic justice check, which are offered as adaptable prompts for professional judgement rather than universal compliance procedures. The article clarifies the framework's limits, including the need for empirical testing, workload‐sensitive implementation, learner‐centred framing and institutional support.