(27 Sept 2025) The attached article discusses the recent significant reduction in the availability of abstracts for scholarly articles from major publishers like Elsevier and Springer Nature on open academic metadata aggregators such as OpenAlex. Over recent years, AI-powered academic search tools have heavily relied on open abstracts for effective research discovery.
However, since November 2024, these publishers have increasingly restricted access to abstracts for non-open access articles, drastically lowering abstract coverage—from around 80% previously to below 40% and as low as 22.5% for Elsevier articles from 2022-2024 in OpenAlex. This change is due to publishers recognizing the value of abstracts for AI training and discovery and negotiating licensing deals, effectively ending the previously “free ride” these AI tools enjoyed. The implications are wide-ranging: AI research search tools will have reduced effectiveness, normal keyword searches in open aggregators will miss many closed-access articles, and bibliometric research will be hampered by the loss of open metadata. Although LLMs could theoretically scrape abstracts directly from publisher websites, this is less efficient and potentially blocked in the future.
In this article, Aaron Tay calls for advocacy from libraries and researchers to push for genuinely open metadata policies, educating users about these limitations, and increasing Open Access rates as a longer-term solution to preserve discovery capabilities in AI research tools.
