Ban for Authors Submitting AI Content 'Welcome but Unenforceable' (www.insidehighered.com)

arXiv, a leading preprint platform, has implemented a one-year ban on authors whose submissions contain "hallucinated" references generated by AI, a move applauded by research integrity advocates. This policy aims to address the growing concern of low-quality and misleading submissions, particularly in fields like mathematics, physics, and computing. Thomas Dietterich, chair of arXiv’s computing section, highlighted that detectable signs of AI-generated content, such as phrases left unedited or inauthentic references, would lead to immediate penalties, requiring future submissions to be validated by reputable peer-reviewed venues. While the initiative is welcomed as a countermeasure against detrimental practices like paper mills, its enforceability raises significant questions. Critics, including postdoctoral fellow Reese Richardson, stress the challenge of policing the volume of submissions, as arXiv may lack the resources to effectively manage bans and appeals amidst a surge in dubious entries, which could include an estimated decline of thousands of manuscripts with false references. This situation underscores a pivotal moment for preprint platforms like arXiv regarding their role in maintaining research quality while still promoting rapid scientific communication. The debate over how to balance these responsibilities is expected to intensify as the landscape of scholarly communication continues to evolve.