AI slop is causing increasing concern everywhere and education is no exception. The number of requests I have received to review submissions to education journals has more than trebled this year. According to gossip the EU education programmes have never before received so many applications and they are likely to change application procedures next year. AI makes it easier but there are downsides.

The open-access preprint repository arXiv has recently announced a strict new policy aimed at curbing the influx of low-quality, AI-generated research papers. The platform, which has become a primary site for circulating research in fields like computer science and mathematics, will now impose a one-year ban on authors who submit papers containing “incontrovertible evidence” of unchecked artificial intelligence generation [1]

While arXiv is most closely associated with the physical and computer sciences, this development sends a clear signal across the academic spectrum, including educational research and vocational education and training (VET). As generative AI tools become embedded in academic workflows, the line between legitimate assistance and intellectual abdication is being drawn in real time.

The Trigger: Hallucinated References and Lazy Prompts

The catalyst for this decisive action is the growing prevalence of obvious AI errors in submitted manuscripts. Thomas Dietterich, a computer scientist at Oregon State University and chair of arXiv’s computing section, highlighted that evidence of unchecked AI use includes “hallucinated references” and LLM meta-comments accidentally left in the text, such as “here is a 200 word summary; would you like me to make any changes?” [2].

Dietterich’s rationale is straightforward: “If a submission contains incontrovertible evidence that the authors did not check the results of LLM generation, this means we can’t trust anything in the paper” [1].

Crucially, arXiv is not banning the use of AI tools for tasks like literature reviews or language editing. Instead, the policy enforces the principle of authorial responsibility. Researchers remain fully accountable for their submissions, “irrespective of how the contents are generated” [1]. If an author copies and pastes biased content, plagiarized text, or fabricated citations from an LLM, the responsibility—and the penalty—rests entirely with them.

The Enforcement Dilemma

The academic community’s response has been mixed, balancing support for the principle with scepticism about its practical application. Research integrity campaigners have largely welcomed the move as a necessary countermeasure against paper mills—organizations that churn out low-quality or fraudulent research for profit [2].

However, the sheer volume of submissions makes enforcement a daunting prospect. In March 2026 alone, arXiv received more than 30,000 submissions [2]. Reese Richardson, a postdoctoral research fellow focusing on research integrity at Northwestern University, questioned the scalability of the ban. “Zhao et al. estimate that thousands of manuscripts containing hallucinated references will be posted on arXiv every year,” Richardson noted. “Does arXiv plan to apply bans for all of these submissions? Issuing these bans requires arXiv staff to adjudicate each case as well as respond to appeals, which I imagine will wind up being quite onerous” [2].

There is a genuine concern that if the policy is only selectively enforced, it may fail to provide a sufficient deterrent against the submission of unverified, AI-generated content.

Implications for Educational Research

For researchers and practitioners in educational research and VET, the arXiv policy serves as a timely warning about the changing norms of academic publishing in the AI era. The pressure to publish quickly can make the drafting capabilities of LLMs highly tempting. However, the reputational risk of being caught submitting unchecked AI output is now being formalised into hard sanctions.

In educational research, where literature reviews often synthesize complex, qualitative pedagogical studies, the risk of an LLM “hallucinating” a citation or misrepresenting a foundational theory is high. The arXiv ban underscores that the core academic competency is no longer just the ability to generate text, but the verification of the information that text contains.

Furthermore, this development touches on the broader debate about the role of preprint servers. Platforms like arXiv were designed to share research quickly, without the traditional gatekeeping of formal peer review [2]. Yet, as AI makes the production of plausible-looking research cheaper and faster than ever, these open platforms are being forced to adopt gatekeeping mechanisms simply to maintain a baseline of trust.

As VET institutions develop their own policies for the ethical use of AI by both students and staff, the arXiv approach offers a clear model: use the tools if they help, but the human author must always take the final responsibility for the output.

References

[1] Ha, A. (2026). Research repository ArXiv will ban authors for a year if they let AI do all the work. TechCrunch. https://techcrunch.com/2026/05/16/research-repository-arxiv-will-ban-authors-for-a-year-if-they-let-ai-do-all-the-work/

[2] Grove, J. (2026). arXiv’s ban for authors submitting AI content ‘welcome but unenforceable’. Times Higher Education. https://www.timeshighereducation.com/news/ban-authors-submitting-ai-content-welcome-unenforceable

About the Image

A man sits thinking and scribbling something in a notebook, with thought bubbles showing indefinite strokes and then a lightbulb. A computer sits in the distance with an LLM chat interface on the screen. The image explores the impact of LLMs on an individual’s writing process. Originally inspired by the concept of “distant writing” developed by Floridi (2025), the image shows how some individuals prompt LLMs with ideas and concepts, acting as a designer, and LLMs provide textual outputs. On the other hand, the image also visualises the distance that some authors put between themselves and the use of LLMs for the purposes of writing, creating, and researching. The image was created by drawing with the program Krita.