Academics used ChatGPT to write peer review reports for Australian Research Council grant applications, with some reports containing telltale signs like 'regenerate response' text and generic summaries without substantive critique.
The Australian Research Council (ARC) received reports that some peer reviewers were using ChatGPT to write assessment reports for Discovery Projects grant applications worth up to $500,000. The ARC_Tracker Twitter account documented instances where researchers received assessor reports that appeared to be AI-generated, including one report that contained the text 'regenerate response' - a clear indicator of ChatGPT use. The reports were described as generic regurgitations of grant proposals without meaningful critique or assessment. Researchers who tested this by inputting their own proposals into ChatGPT received very similar summaries to what they had received as official assessments. The incident prompted the ARC to issue a statement warning that using AI tools for peer review may constitute a breach of confidentiality obligations, as assessors are required to keep grant proposal information confidential. The underlying causes included excessive workloads for academics who must review up to 20 proposals of 50-100 pages each, often without dedicated time allocated in their university workload models, and the absence of clear ARC policies explicitly prohibiting AI use in peer review processes.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
Using AI systems to gain a personal advantage over others such as through cheating, fraud, scams, blackmail or targeted manipulation of beliefs or behavior. Examples include AI-facilitated plagiarism for research or education, impersonating a trusted or fake individual for illegitimate financial benefit, or creating humiliating or sexual imagery.
Human
Due to a decision or action made by humans
Intentional
Due to an expected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed