Copy article

New AI Ethics Book Under Fire For 'Faked' AI Journal Citations

ended 15. December 2025

The Times reports  Springer Nature faces criticism after a newly released book on AI ethics was found to contain fabricated citations, non-existent journals, and made-up research papers.

Social, Ethical and Legal Aspects of Generative AI is marketed as an authoritative review of ethical dilemmas in artificial intelligence but includes references to journals and papers that do not appear to exist.

Investigations by The Times and academic experts revealed that in some chapters, more than 70% of citations could not be verified. One chapter contained eight citations that could not be traced, while another cited a paper allegedly published in the "Harvard AI Journal," a publication that doesn't exist. 

Guillaume Cabanac, a computer science professor at the University of Toulouse who specialises in detecting fraudulent academic work, described the findings as "research misconduct: falsification and fabrication of references.

Dr Nathan Camp of New Mexico State University conducted a separate review and found multiple erroneous, mismatched, or entirely invented references. In some cases, details from different genuine papers appeared to have been combined. Camp noted that although it is difficult to prove definitively that AI generated the false citations, their pattern strongly suggests AI involvement. 

Springer Nature has acknowledged the concerns. James Finlay, vice-president for applied sciences books, said the publisher's research integrity team is investigating the issue as a priority. He stressed that while rigorous checks are in place to prevent errors, a small number of issues may occasionally slip through the peer-review and editorial process.

The controversy comes amid growing scrutiny of AI's role in academic publishing. This is the second Springer Nature book pulled this year for the same reason. The Bookseller reported the publisher reworked Mastering Machine Learning: From Basics to Advanced after discovering multiple fictitious references.

We want your views:

  • How do we rebuild trust in traditional publishing when fabricated citations pass editorial review?
  • What does it say about humans when a book about AI ethics contains AI-generated fabrications?
  • If peer review can't solve this problem reliably, what quality assurance mechanisms actually work for AI-assisted publishing?
  • What would responsible AI deployment look like in academic publishing where accuracy isn't optional?

 

3 responses from the Newspage community

Copy all

Star Quote
Copy

Colette MasonVerified
AI Ethics Consultant at Clever Clogs AI
This is a masterclass in what happens when you treat AI like a shortcut instead of a tool that needs oversight.

This is "set and forget" AI culture coming home to roost. Someone appears to have fed prompts into an LLM, trusted the output without verification, and Springer's rigorous checks failed to catch references to journals that don't exist. This is a failure of human accountability.

Academic publishing trades on credibility. Once that's gone, it's gone. You can't rebuild trust by calling fabrications occasional errors or hiding behind "AI involvement is hard to prove." You rebuild it by admitting the process failed, implementing human verification at key stages, and making AI disclosure less stigmatised so assistance becomes visible and auditable.

The real question isn't whether AI can generate convincing fabrications. It's whether we're willing to build systems that check the work. Because if you can't trace it, you can't trust it, and you shouldn't be publishing it.
Copy

Patricia McGirrVerified
Founder at Repossession Rescue Network
AI does not lose control when it is supervised properly. It does so when production is prioritised over protection. A book on AI ethics packed with fabricated citations is not ironic, it is a governance failure. Human oversight was either rushed, diluted or treated as a formality. That is how invented journals pass editorial review and authority collapses in public. Publishing runs on trust. If citations cannot be relied on, nothing else in the system holds. Peer review cannot compensate for volume and speed that exceed human checking capacity. Responsible AI use means fewer outputs, harder checks and named human accountability that cannot be automated away. When protection is seen as friction, reputation becomes the cost.
Copy

Rohit Parmar-MistryVerified
Founder at Pattrn Data
You couldn’t make it up, but clearly, the AI did. The irony of an AI ethics book failing the most basic ethical test, telling the truth, is a plot twist even Philip K. Dick would find a bit too on the nose.

We’ve been sold a sci-fi dream where AI is a super-intelligent oracle, a HAL 9000 for the boardroom. But this scandal proves we’re dealing with something far less impressive: a fiction engine that prioritises plausibility over fact. This is 'automation theatre' at its worst. The problem isn't that the machine 'hallucinated', that is a known feature of the tech. The scandal is that the humans involved were too asleep at the wheel to notice, likely blinded by the hype.

We don't need Asimov's Three Laws of Robotics to fix this; we just need humans to pick up a red pen. If you treat AI like a magic box rather than a tool that needs supervision, you don't get 'intelligence', you get a very confident, very wrong fantasy.