This work presents an interdisciplinary analysis of institutional guidelines for the adoption and use of (generative) AI in higher education. Through an inductive, multidisciplinary, theoretically informed analysis of around forty publicly accessible documents (in English, German, and Spanish) from predominantly European universities, we extracted 30+ exemplary fragments, which function as heuristics that illustrate rhetorical moves and assumptions that surface in these publications. We show how to engage with such texts critically. To do that, we identify nine recurring narrative patterns and discuss four of them in detail. Namely, 1) rhetoric of inevitability and technological determinism; 2) exaggerated narratives that overstate the general capabilities of the technology; 3) spurious comparison to human intelligence or Anthropomorphism; 4) ethics and critical washing, i.e., ethical or critical examination but only superficially and inconsequentially. These patterns, in different ways, serve as narrative gambits that seek to normalize the institutional adoption of AI, bypassing serious critical scrutiny. We develop the notion of “responsible realism” to characterise the stance in which institutions acknowledge the harms and ethical problems of commercial AI yet still promote its use, shifting the burden of mitigation onto individual users by calling to “responsible use.” We conclude by calling the academic community to reject narratives designed to dilute responsibility and obfuscate serious consideration of whether these technologies should be adopted at all, and to reclaim responsibility by critically engaging with guidelines and policies that, even if unintentionally, serve the interests of tech corporations above those of the academic community and society writ large.