An artificial intelligence research report co-authored by Accenture PLC (NYSE: ACN) has been found to contain fabricated citations, misattributed research findings, and case studies that appear not to exist at all, prompting accusations that the consulting firm released a report contaminated by AI hallucinations.
The study, jointly published by the IT consulting company and a Spanish university, examining the potential applications of artificial intelligence in African agriculture, contains numerous factual errors and 15 citation mistakes, including inaccurate paper titles or author names. These problems were uncovered by AI detection firm GPTZero during an investigation.
Edward Tian, co-founder of GPTZero, stated that upon verification, these errors "overwhelmingly match the typical patterns of AI hallucinations rather than the occasional inaccuracies found in human research."
Accenture PLC (NYSE: ACN) was appointed in September as an independent evaluator of the safety of Anthropic's AI models. The company said there is currently "no evidence" that the errors were caused by AI hallucinations, but subsequently stated it would revise the report. Accenture PLC (NYSE: ACN) said: "Our review indicates that the errors identified in the report were caused by human error. There is currently no evidence to support the conclusion that these errors were generated by AI hallucinations." The consulting firm declined to specify the basis for this conclusion and did not further explain how the errors actually arose. Accenture PLC (NYSE: ACN) said the corrected report will be republished, adding: "We always welcome the opportunity to further improve the quality of our reports."
The report, published in November 2025 and titled "Leapfrogging Development: Unlocking Africa's Potential in Global Value Chains with Artificial Intelligence," was jointly released by Accenture PLC (NYSE: ACN) and IE University. Two of the report's four authors are Accenture PLC (NYSE: ACN) employees, and both editors are also from the company.
In recent years, Accenture PLC (NYSE: ACN) has aggressively reshaped its business model, not only making extensive use of artificial intelligence but also providing consulting services to clients on adopting the technology. The report, published on Accenture PLC (NYSE: ACN)'s official website, references technology provided by an organization called SAATN twice, but uses two different names for the organization: "Southern Africa AgriTech Network" and "Southern African Agriculture Transformation Network." In reality, neither of these names appears to correspond to an existing organization.
The report also claims that predictive AI significantly improved the detection of cocoa swollen shoot disease, achieving a detection accuracy of 99% in Colombia and Ecuador. However, the reference cited for this claim is actually an article focused on West Africa, which makes no mention of these two South American countries and provides little evidence that the disease exists there. Other errors include incorrectly listing Ghana and Uganda's rankings in Oxford University's 2023 AI Readiness Index and citing a webpage link that cannot be accessed.
Multiple citation-related errors were also found in the report. Tian stated: "Errors humans make when citing references are usually small-scale and isolated, such as incorrect years, non-standard formatting, or broken links." He added: "In contrast, large language models generate references that sound completely credible but do not actually exist. This leads to fabricated paper titles, plausible-sounding but non-existent organizations, and case studies cobbled together from real institutions and individuals."
Accenture PLC (NYSE: ACN)'s report claims that in Colombia and Honduras, "digital platforms led by young people and women, after incorporating inclusive governance structures, have already achieved high adoption rates." However, both references cited for this claim appear not to exist. The 49-page report also contains several webpage links that cannot redirect to the original articles cited. One citation misspells an author's name and lists two authors who were not actually credited with the research paper.
In recent years, Accenture PLC (NYSE: ACN) has strongly encouraged employees to use artificial intelligence, not only developing related training courses but also incorporating AI tool usage into employee promotion evaluation criteria. Meanwhile, investors are closely watching whether artificial intelligence can increase the company's revenue per employee. A round of layoffs the company conducted last year focused mainly on employees who could not acquire AI skills through retraining.
However, AI detection tools including GPTZero are not infallible and have been criticized for sometimes incorrectly flagging human-written content as AI-generated. GPTZero stated that it manually verified every error its tool flagged in the Accenture PLC (NYSE: ACN) report and conducted additional verification using open-source intelligence. GPTZero had previously also discovered AI hallucination issues in research reports by PwC, KPMG, and EY. KPMG and EY subsequently withdrew their respective reports after the related findings were published.
IE University stated: "The report's authors have always maintained extremely rigorous standards in conducting their research and analysis. If any errors are confirmed, they will make necessary corrections to the report and strengthen existing processes to prevent similar issues from recurring." The university added that the report's authors are re-examining the sources and data underlying the report to understand how these discrepancies arose.
Comments