This study explores the arrival of Generative Artificial Intelligence (GenAI) technologies in a collaborative Problem-Based Learning (PBL) environment from the perspective of students. It examines the emerging practices students develop individually and in groups as well as the tensions they navigate in response to GenAI's sudden arrival, its inherent limitations and multifaceted affordances, as well as the evolving landscape of formal and informal rules that frame its use. Grounded in Cultural-Historical Activity Theory, the study adopts an exploratory research approach, utilizing both qualitative and quantitative data from a survey administered to 1st and 5th semester IT students in a systemic PBL university (n = 83). Quantitative data were analyzed using descriptive statistics and principal component analysis, while qualitative data were analyzed through thematic analysis. The findings indicate a high adoption rate and generally favorable attitudes toward GenAI across both semesters. While 1st semester students were more likely to report no prior experience with GenAI compared to 5th semester students, no notable differences were observed regarding frequency or patterns of use. Overall, the arrival of GenAI manifests in the students' individual and collaborative study practices by enabling or obstructing both epistemic and pragmatic mediations of the objects. However, these interactions have been found to be accompanied by several tensions--specifically between the mediation of different objects, between students and their project groups, and between students and the rules framing GenAI use, which are discussed and contextualized in light of theory and contemporary research.