At a glance: PwC consulting reports contain verifiably false data and fabricated sources indicating AI hallucinations.
Several reports from consulting firm PwC contain erroneous data and fictitious source citations that result from AI hallucinations. The findings were documented by GPTZero.
Consulting firm PwC has published data in several reports that does not match the facts. Upon verification with AI detection service GPTZero, both factually incorrect information and fabricated source citations were identified, pointing to AI generation.
The errors in PwC’s reports reveal a central risk in integrating large language models into the creation of business documents: AI models tend to engage in so-called hallucinations, in which they produce plausible-sounding but factually false information and invented references. This is particularly critical when such reports are used as the basis for business or regulatory decisions.
For Chief Data Officers, this presents a dual governance challenge: on one hand, processes must ensure that AI-generated or AI-assisted content is verified before publication. On the other hand, the case underscores the need to explicitly flag to stakeholders (internal leadership, customers, regulators) where AI was deployed in document creation, in order to handle trust and liability issues transparently.
Source: www.golem.de · Published 30 July 2026
Lumi AI News — AI-assisted curation pursuant to Art. 50 EU AI Act. Paraphrasing and classification by Lumi News Pipeline v1.7.3.