Kudos releases Phase I report from “Taming the Crocodile” project and invites new sponsors to join for Phase II.
For immediate release – Oxford, UK – July 15th, 2026 – New research led by Kudos reveals that AI has already captured a third of scholarly information-seeking, with many researchers unaware of the extent to which their view of the scientific literature is now AI-mediated. The heavily anticipated Phase I report from ‘Taming the Crocodile’ has been delivered to sponsors, and distils findings, analysis and recommendations from two surveys (11,500 researchers and 200 librarians) and an environment scan of more than 300 sources. A comprehensive list of actions sets out priorities for marketing, editorial, customer service, account management and technology teams, as well as a roadmap for the longer-term strategic response.
Headlines from the Phase I research include:
- A third of researchers are starting their information seeking with AI. One in ten researchers turn directly to AI tools as the starting point for their information seeking; a further 25% start in general search engines where AI-generated answers are now embedded. This extraordinarily rapid shift is even more pronounced among early-career and corporate researchers (twice as many start with an AI tool).
- Alarmingly, 40% wrongly think AI overviews generated on the fly in search engines have been created or curated by a human. That figure was higher among students (46%). Librarians report users routinely requesting articles that do not exist, because of hallucinated responses by AI.
- Usage is dropping, and putting subscriptions at risk. Librarians are noticing the effect of AI discovery on usage of primary sources and are more likely to have seen this as a negative effect (drop in usage rather than increase). While they agree that a decline in usage doesn’t mean a decline in value, usage metrics are still driving purchasing behaviours.
- Librarians want publishers to play an active role in ensuring AI systems accurately represent scholarly content; they want transparency about publishers’ engagement with AI, and have insightful views on how content should be licensed for AI training or agentic querying.
- The single most frequent recommendation from AI optimization agencies is to develop an AI discovery layer of structured, "answer-ready" plain-language summary content ("bottom line up front" – what is this about and why is it important). What was a “nice to have” is now essential if you want to reclaim / rebuild traffic, and retain brand visibility and citations.
The Phase I report provides immediate guidance on optimizing content for AI; Phase II of the study will work with the sector to develop best practice around schema and mark up for AI optimization. Wider recommendations include:
- A step-by-step process for baselining and tracking AI visibility.
- Key topics to cover in educational materials for librarians and end users.
- Latest guidance on generative engine optimization (GEO) and AI optimization for search engines (AIO) including server / crawler settings, coding “do’s and don’ts” for AI readability, visibility in knowledge graphs and indexes.
- Custom analysis and recommendations for maintaining visibility via AI in China
- Pivoting publishing value proposition towards curation and stewardship of knowledge (as opposed to distribution) including cryptographic attestation of trust.
"The findings give us all such a strong foundation to help adapt our industry infrastructure," says Emilie Delquié, Chief Product & Customer Success Officer at Silverchair, which sponsored the study.
"This is the first evidence base we have seen that connects how researchers actually search to what authors and publishers should do about it," said Shane Rydquist, Associate Vice President, Delivery and Solutions at CACTUS Communications. "The finding that matters most is that researchers rely on AI but do not trust it, and that evidence and clear attribution are what earn both the click and the citation. We are already using it to shape how we help authors make their work discoverable and credible in an AI-mediated world."
Dan Penny, Head of Market Intelligence at Springer Nature, adds, "The fast-moving and unpredictable impact of AI on all aspects of scholarly communication makes it essential to stay ahead. We all need to act quickly, and responsibly, on the findings here to ensure that we continue to support researchers, the wider community, and the integrity of scholarly communication."
The study's other headline sponsors were Elsevier, Emerald and the Royal Society of Chemistry; there were 18 sponsors in total.
The full report can be bought by itself or as part of a package including participation in Phase II, which will move from diagnosis to coordinated response: benchmarking how AI-driven discovery is affecting publishers' usage and revenues; observing researcher behaviour directly to meet and shape changing expectations; and developing shared standards for the content, mark-up and machine-readable signals the sector wants AI systems to ingest, interpret and attribute. For more information, visit https://info.growkudos.com/croc-pub
— ENDS —
Contact:
Charlie Rapple
charlie@growkudos.com
About Kudos
Kudos (growkudos.com) is the platform for showcasing research, helping more people find, understand and use it. Over 650,000 researchers use Kudos to bring science to life through the power of storytelling. Kudos is the only toolkit designed specifically to drive research communications, engagement and impact, helping researchers cross boundaries, achieve more influence, and gain more recognition.
