Argus: Academic Integrity in the Era of Generative AI

arXiv:2609.36073v1 Announce Type: new Abstract: The rapid proliferation of large language models (LLMs) in the context of education has introduced significant challenges in enforcement of academic integrity, especially in programming courses. We present Argus, an automated detection system for LLM-assisted student work in undergraduate C programming assignments. Argus integrates behavioral and stylistic indicators to create a holistic picture of the student's progress through an assignment and surfaces anomalies that point to potential misuse of LLM assistance. We quantify and analyze data over six years of Spring semester offerings in a large-enrollment CS2 course at Purdue University using Argus, finding that 45% of enrolled students exhibited patterns consistent with LLM-assisted code development in Spring 2026. To contextualize these results, we analyze the relationship between flagged LLM use and student performance on written, in-person proctored examinations, and find a significant negative correlation. We also explore the problem of mitigating false positives, recognizing that erroneous accusations of academic integrity carry significant consequences for students and instructors alike, particularly in the context of large enrollment courses. We argue that any automated detection system must be accompanied by a structured process for human review. We discuss the consequences for future course design, changing academic policy as these tools become more ubiquitous, and the pedagogical implications of LLM-based tools in computer science education.

LLM use negatively correlates with student performance

This documents rising LLM use among students correlating with weaker fundamental skills on proctored exams, a concerning signal for education. But it also shows institutions actively building detection and review infrastructure, a partial mitigating response. Net implication is moderately negative but contained to an academic integrity domain.

The study found that 45% of students in a large CS course used LLMs for code development in 2026, and this usage was significantly negatively correlated with performance on proctored exams. This suggests a material detriment to student learning and academic integrity in higher education, directly impacting human flourishing through reduced skill acquisition.

The study finds widespread use of LLMs in academic work, with a negative correlation to exam performance. This suggests a material concern for educational integrity and student learning outcomes.

SOURCE     arXiv cs.CY (Computers and Society) (tier 1)
   ↓
DOCUMENT   6bd7ea4e-e59b-498f-8e53-9ea86ec7c092
           https://arxiv.org/abs/2609.36073
   ↓
EVIDENCE   2 extracted excerpts
   ↓
MODEL RUN  4 runs, methodology 1.0.1
   ↓
SCORE      -60  (Adverse)
   ↓
CONFIDENCE 80%