Academic Fraud: Google Scholar Citation Gaming Exposed
In February 2025, researchers at NYU Abu Dhabi demonstrated how easily the academic record can be compromised.
After building a fake academic profile and uploading ChatGPT-generated papers to preprint servers, they paid $300 via WhatsApp to have 50 citations delivered. While the fraudulent Google Scholar citation count spiked immediately, the most concerning discovery was that the citations persisted even after the source papers were deleted.
This is not a newly discovered vulnerability. Rather, it reflects how the platform has functioned for fifteen years without structural change.
What $300 buys you
Historical experiments confirm this lack of oversight. In 2012, Universidad de Granada researchers generated 774 citations from six fake documents.
H-index scores rose, and younger researchers saw their metrics multiply sixfold. These documents remain discoverable for years because Google Scholar lacks a mechanism to exercise the “right to be forgotten” regarding fraudulent data.
A recent blog finding by Reese Richardson highlights a thriving black market where 100 citations are sold for $300. While Google acted quickly to remove citations for “Larry the cat”. A fictional researcher created to mock the system purchased citations documented a year prior that remain indexed. Data suggests that 32% of anomalous citations no longer exist on their original servers.
Yet every Google Scholar citation earned through these deleted papers remains on the platform. Leaving metrics intact despite the disappearance of evidence.
Citation cartels: Displacing Stanford and Princeton
The institutional impact of these practices is starkly illustrated by a shift in global mathematics rankings.
A finding by Docampo, published in Science Analysis, found that UCLA and Princeton dominated the list of highly cited mathematics papers between 2008 and 2010. However, by the 2021-2023 window, China Medical University in Taiwan led the list with 95 highly cited papers. Up from zero just a decade earlier. In that same timeframe, UCLA’s count dropped to a single entry.
These metrics were not driven by a breakthrough in research. But by researchers at the same institutions citing one another within predatory journals. In his Chronicle of Higher Education analysis, Docampo described the phenomenon bluntly:
“People published in journals that no serious mathematician reads, whose work was cited by articles no serious mathematicians would read, from institutions nobody knows in mathematics.”
The scale of this manipulation eventually forced a drastic industry response. Clarivate removed the entire field of mathematics from its Highly Cited Researchers (HCR) list. This decision was not based on a precise identification of every fraudulent entry.
Rather, the manipulation had become so pervasive that legitimate mathematical achievements became indistinguishable from gamed metrics.
The AI feedback loop: A new threat
A new and dangerous frontier emerged in early 2026 with the integration of AI-generated search summaries.
As Google’s AI Overview feature began citing indexed material, it inadvertently surfaced articles from fake publication venues like JSTAR and IJEIMS. When a fraudulent Google Scholar citation is used to train or inform an LLM, the misinformation is “laundered” through an authoritative AI interface.
This creates a feedback loop where AI-generated fake papers produce citations that then inform the very AI models used by students and professionals. If the underlying index is compromised, the generative layers built on top of it will inevitably hallucinate a “consensus” that does not exist in legitimate peer-reviewed literature.
The predatory journal ecosystem
The volume-over-quality model fuels the crisis. In 2023, MDPI Mathematics published over 4,700 articles, while the prestigious Annals of Mathematics published just 22.
Both carry Web of Science indexing, which creates an illusion of identical institutional weight. Because predatory publishers generate revenue through Article Processing Charges (APCs), they have an incentive to maximize volume, often at the expense of peer-review rigor.
Manipulation methods and platform gaps
The system has five documented methods for categorizing its vulnerabilities, each exploiting a specific gap in the way Google Scholar citations are verified.
| Manipulation Method | Detection Difficulty | Platform Response |
|---|---|---|
| Citation Purchasing | Low (Sting operations) | No systematic action; reactive only |
| Preprint Deletion | Very High | Citations persist after source removal |
| Self-Citation | Moderate | No exclusion mechanism provided |
| Citation Cartels | High (Ambiguous patterns) | Field-wide removal from lists (Clarivate) |
| Fake Author Networks | Moderate | Reactive removal when reported |
Universities created the market
As Cameron Neylon noted in Science analysis, “The stakes are high, movements in rankings can cost or make universities tens of millions of dollars. It is inevitable that people will bend and break the rules to improve their standing.”
Hiring committees that base their decisions on a candidate’s Google Scholar citation count or H-index as a primary signal are using metrics that individuals can purchase for $300.
This crisis is a textbook example of Goodhart’s Law: when a measure becomes a target, it ceases to be a good measure. Citations once measured genuine research impact, but once universities transformed them into a hiring target, gaming the system became inevitable. Consequently, the IMU and ICIAM issued joint recommendations in September 2025, urging institutions to move away from gameable metrics, decouple funding from bibliometrics, and prioritize reading actual publications over simple counts.
Google’s abdication
The platform’s history is a fifteen-year timeline of documented failures. From Ike Antkare’s fabricated papers in 2010 to Cyril Labbé’s use of 102 auto-generated papers to create a fictional top-tier scientist, the vulnerabilities have remained open. Even after Larry the cat was removed in 2024, purchased citations documented by researchers like Richardson remained indexed.
Google’s response has been consistently reactive, removing individual cases only when they become embarrassing, while implementing no systematic fixes. This lack of integrity extends to features like AI Overviews, which Richardson found to be citing fake publication venues such as JSTAR and IJEIMS.
As long as citations from deleted papers persist and purchased metrics stay indexed, the platform remains a primary conduit for academic fraud.
What IT procurement should actually check
When evaluating vendor credentials or technical staff, the “tell” isn’t the count, but the source. A journal publishing thousands of articles with a massive editorial board is a different entity from a specialized journal publishing two dozen.
If a profile shows a sudden Google Scholar citation spike from the researcher’s own institution, it mirrors the cartel pattern Docampo documented. If papers exist on institutional sites but are missing from the preprint servers that supposedly hosted them, it suggests the deletion tactic documented by Nature.
China Medical University’s leap from zero to 95 math papers while Princeton dropped to one isn’t a shift in leadership; it is manipulation at scale.
Distilled
- The mechanism: Ibrahim et al. purchased 50 citations for $300; the papers were deleted, but the citations stayed.
- The displacement: Citation cartels displaced Princeton and UCLA in math rankings, leading Clarivate to remove the entire field from its HCR list.
- The industry problem: MDPI Mathematics published 4,763 articles in 2023, while Annals of Mathematics published 22. Both share the same indexing legitimacy.
- The platform failure: Across fifteen years, Google has offered zero systematic fixes. Allowing the platform to remain the primary tool for faking academic prestige.
For IT procurement and hiring, a verification framework isn’t paranoia. It is basic due diligence for a metric that costs less than a smartphone to fake.
