Generative AI and the Risk of Algorithmic Compliance in University Academic Integrity Policies

Generative artificial intelligence has made the relationship between academic integrity and institutional verification more difficult to settle. A declaration, an oral discussion, or a record of drafts can support learning, but it can also become a demand that students continually demonstrate their legitimacy. This article develops the concept of algorithmic compliance pedagogy through a critical reading of ten public documents from six universities in the United Kingdom, continental Europe, and North America, retrieved on 23 October 2026. The analysis combines reflexive thematic reading with attention to policy wording, the allocation of responsibility, and the grounds on which authorship is questioned. The documents do not support a uniform movement toward automated surveillance. Several explicitly reject AI detectors, while preserving disclosure requirements, authorship interviews, and instructor judgments about writing. Five themes examine conditional permission, restrictions on detection, professional discretion, documentary authenticity, and the relationship between compliance and educational purpose. The central concern is that suspicion can migrate from a detector score to judgments about style or the performance of a writing process. This is a policy risk rather than an observed effect on students. The article distinguishes necessary assessment assurance from disproportionate verification through its connection to learning, the burden of documentation, institutional responsibility, and opportunities to contest an inference. It argues that protecting standards requires universities to account for the grounds of suspicion as well as requiring students to account for their work.