Recent advancements in Artificial Intelligence (AI), particularly AI chatbots like ChatGPT, have sparked significant discussions in higher education regarding their impact on teaching and academic integrity. This study examines the integration of ChatGPT into a philosophy of education course for first-year Education students at a Swedish university. Using a sociocultural approach and the concept of mediating devices, the study investigates how ChatGPT can be effectively integrated into higher education to support academic literacy and the challenges and opportunities it presents. The empirical material was collected through semi-structured focus group interviews, observational notes, and students' assignments. The findings suggest that when used intentionally with clear pedagogical goals, ChatGPT can support academic literacy development. Its ability to generate coherent text, summarise complex concepts, and provide initial structures for academic writing proved particularly beneficial for first-year students. However, limitations such as inaccuracies and lack of nuance provided opportunities for students to engage critically with the generated material and develop strategies to address these shortcomings. The findings highlight that both the teacher's guidance and the social aspects of learning, such as group discussions and collaborative work, were essential in leveraging ChatGPT's strengths and addressing its limitations. Therefore, this paper advocates for a balanced approach that emphasises pedagogical grounding and responsible AI integration to ensure the ethical and effective use of AI tools. Understanding the interplay between ChatGPT's integration and the social dynamics of teaching is crucial for assessing its impact on students' learning experiences.