How AI-vulnerable is Australian higher education assessment? A computational audit of Group of Eight Arts and Humanities units, 2022–2026

Generative AI (GenAI) is profoundly disrupting higher education, prompting educators to redesign course and assessment practices. Yet evidence of the potential impact of GenAI on assessments and subsequent module redesigns remains unclear. Through data scraping and classification, this article audits 15,587 annual Arts and Humanities unit records and 53,915 assessment items across Australia’s Group of Eight universities from 2022 to 2026. This article presents the AI Vulnerability Index (AIVI) which rates assessment exposure against relative GenAI capabilities in each year. At all seven institutions with comparable data, average exposure rose sharply between 2022 and 2024 and remained above 2022 levels in 2026, although subsequent trends differed between institutions. In 2026, very-high-exposure items accounted for 53.1% to 70.7% of allocated marks, while 53.1% to 79.1% of units scored at least 0.60. Exposure was concentrated in unsupervised writing, especially take-home essays and research reports. Institutional policy categories showed no clear pattern of lower exposure. High exposure, however, is not intrinsically undesirable. Where GenAI use is authorised and aligned with learning outcomes, it may offer a pedagogical opportunity. Governance should therefore examine the mix of assessments, the intended role of GenAI, and evidence of learning required of students.