The growing use of AI-based text generation tools such as ChatGPT is reshaping academic practices and raising new questions about academic integrity. As higher education institutions grapple with policy, pedagogy, and ethical boundaries, students and teachers are left navigating unclear expectations. This thesis explores how Swedish university students and teachers perceive generative AI's impact on academic integrity, particularly concerning plagiarism and responsible use. A targeted concept-centric literature review was conducted to map existing research on AI in education, academic ethics, and evolving pedagogical responses. The findings reveal conflicting perspectives, conceptual ambiguity, and a lack of empirical insight into how academic communities engage with these tools. The thesis applies a qualitative research design to address this gap, using purposive sampling and semi-structured interviews with students and teachers across disciplines. Thematic analysis identifies six themes: informal norms, ethical uncertainty, and shifting perceptions of authorship. The findings show how academic norms are being renegotiated in practice and may support the development of clearer institutional strategies and ethical frameworks for AI use in higher education.