The rapid development of Artificial Intelligence (AI) opens up new didactic possibilities for fostering disciplinary language competence in the natural sciences. AI-supported tools enable personalised access to disciplinary texts, promote self-directed and cooperative learning, and facilitate heterogeneous learning groups’ access to specialised terminology, which is often perceived as a “new language.” Using practice-oriented classroom scenarios, the article illustrates how learners analyse conceptually demanding passages, develop targeted prompts for AI systems, and critically evaluate the generated output. This approach supports both reflection on disciplinary language structures and metacognitive work, and strengthens learners’ ability to engage in scientifically grounded thinking. Research findings underline the need for a didactically reflected integration of AI in order to ensure disciplinary accuracy and communicative competence in scientific contexts. Key challenges include safeguarding content accuracy and raising awareness of risks such as declining learner agency and increasing automation effects. The article discusses the opportunities and risks of AI-supported disciplinary language learning and offers practice-oriented suggestions for responsible use that sustainably strengthens scientific communication in a media-literate and collaborative way.