Global consultancy firm Accenture is facing scrutiny after an analysis by AI-detection software provider GPTZero alleged that several of its published research papers contain fabricated case studies and fictitious citations typical of generative artificial intelligence hallucinations.

Flawed Citations in Policy Research

The primary document cited in the investigation is titled “Leapfrogging Development: Unlocking Africa’s Potential in Global Value Chains with Artificial Intelligence,” a policy paper co-authored by Accenture in collaboration with IE University in Madrid. According to GPTZero, that single publication features 15 specific inaccuracies characterized as AI hallucinations, alongside four additional flawed citations. The identified issues involve references to sources that do not exist, materials published under completely different titles, or cited works that fail to support the factual claims made in the text.

The policy paper was not an isolated case in the analysis. In total, GPTZero evaluated three Accenture-linked documents, identifying a combined 31 hallucinated references, numerous unsupported assertions, and hallmarks indicative of machine-generated text.

Detection Findings and Risks to Research

GPTZero stated that it identified the anomalies by running the documents through an automated pipeline equipped with its Hallucination Detector. The organization emphasized that each flagged hallucination was subsequently reviewed and verified by a human analyst to prevent automated false positives.

The AI-detection publisher raised broader concerns regarding the downstream impact of fabricated references entering the public domain from prominent corporate entities. The group contended that hallucinated citations present a clear danger to researchers, academics, consultants, students, and general information seekers. According to GPTZero, unchecked false claims published by reputable institutions can be repeatedly cited and amplified across the internet, warning that over time such recycled errors risk poisoning the broader information ecosystem even if human readers never consult the original source papers.

GPTZero has previously examined citation discrepancies across corporate publications, having pointed out inaccurate and false claims in reports issued by auditing firm PwC.

Accenture Denies AI Generation

In response to the findings, an Accenture spokesperson speaking to The Telegraph denied that generative AI tools were responsible for the fabricated citations, asserting that the discrepancies were the result of human error rather than machine hallucinations.

The spokesperson stated that there is no evidence indicating the errors were produced via AI hallucination, adding that the document is currently undergoing corrections ahead of a planned re-release. The company maintained that it welcomes opportunities to strengthen the quality and rigor of its published findings.

The controversy surfaces as Accenture continues to urge its global workforce to integrate artificial intelligence into day-to-day operations and business workflows. While the examined paper was produced as an external collaboration outside Accenture’s standard internal publication pipeline, the scrutiny highlights the challenges consultancy firms encounter in verifying data integrity and maintaining editorial standards in an era of rapid AI adoption.