Thought Leadership
According to GPTZero, a report on AI in Africa’s agriculture contains 15 false citations. Accenture speaks of human error.
Consulting firm Accenture is facing criticism over a research report containing numerous false citations, misattributed studies, and apparently non-existent case studies. The Financial Times reported on this. The report is titled “Leapfrogging Development: Unlocking Africa’s Potential in Global Value Chains with Artificial Intelligence” and examines how AI could increase the productivity of agriculture in Africa. Accenture had published it in November 2025 jointly with IE University in Madrid. Two of the four authors and both editors are Accenture employees.
15 out of 89 Sources Affected
GPTZero, a company specializing in detecting AI-generated text, examined the report using an automated procedure. Out of 89 citations, it classifies 15 as AI hallucinations, meaning references to sources that do not exist, were published under a different name, or do not support the statements. Four additional citations are said to be flawed. According to GPTZero co-founder Edward Tian, the errors predominantly match typical patterns of AI hallucinations, such as plausible-looking but non-existent sources and fabricated case studies.
GPTZero also examined two technical papers by Accenture researchers that were published on the arXiv platform in May 2026. Across all three papers, the company found 31 hallucinated citations out of a total of 245 sources. According to GPTZero, a citation regarding the malicious chatbot FraudGPT pointed to a fake GitHub repository containing real malware.
Accenture Wants to Correct Report
Accenture denies that AI is responsible for the errors.
“Our review shows that the errors identified in the report are due to human error. There is no evidence supporting the conclusion that they resulted from AI hallucinations.”
Accenture
Accenture did not disclose how it arrived at this conclusion. The report is to be corrected and republished.
Accenture is heavily pushing the use of AI within its own organization. CEO Julie Sweet had announced in 2025 that the firm would part ways with consultants who do not use AI. Most recently, according to reports, the group capped the centrally funded budget for Claude at 29 US dollars per person and month. Similar allegations were made against Deloitte in 2025. That company had to refund part of its fee to the Australian government for a report that contained flawed citations created with the help of AI.
(Editorial Team)
