Accenture co-authored an AI report about African agriculture that contained false citations, incorrectly attributed research and non-existent case studies, prompting accusations of AI hallucinations. An investigation by GPT Zero identified 15 citations with errors including incorrect titles and author names. AI detection co-founder Edward Tian stated the errors were overwhelmingly consistent with AI hallucination patterns rather than typical human research inaccuracies.
Accenture initially claimed no evidence the errors stemmed from AI, but later pledged corrections, stating the mistakes resulted from human error. The November 2025 report, titled "Leapfrogging Development: Unlocking Africa's Potential in Global Value Chains with Artificial Intelligence," was published alongside IE University. The paper references two non-existent networks -- the Southern Africa Agri Tech Network and Southern African Agriculture Transformation Network.
It also claimed predictive AI improved swollen shoot disease detection in cocoa plants to 99 percent accuracy in Colombia and Ecuador, though the cited source focused on west Africa. Additional errors included incorrect rankings in a 2023 Oxford AI readiness index and broken web links. The 49-page report further contained citations with non-existent authors and case studies stitched to real actors.
Accenture has heavily encouraged staff AI use for promotions amid investor scrutiny.
Source: Financial Times Companies · Summarized by HeadlinesBriefing