Consultancy firm KPMG has made an embarrassing blunder. A report intended to demonstrate the benefits of AI turned out to be full of AI hallucinations.
KPMG has pulled a report from its website after it was found to be full of errors. In the report, ironically titled Redefining excellence in the age of agentic AI, KPMG explained how several organizations in Switzerland and the UK were deploying AI. However, some use cases turned out to be total fabrications, the Financial Times reports, based on research by GPTZero.
For instance, the report claimed that the Swiss bank UBS uses AI agents for investor advice, risk management, and compliance monitoring via a platform developed with Microsoft. ‘Factually incorrect,’ the bank told the Financial Times. In the case of Manchester’s public healthcare system, the report had misinterpreted a press release.
KPMG has issued a mea culpa, stating in a response that it ‘takes the accuracy and integrity of its published content seriously.’ “We expect all our employees to adhere to our guidelines for the responsible use of AI, which include human oversight to validate content and check independent sources,” KPMG added.
Caught out
Yet KPMG is not the first major consultancy firm to be caught out by the careless use of AI. Last year, Deloitte was caught using fabricated quotes and references to non-existent sources in a report for the Australian government, for which the government had paid $290,000. The Canadian branch of EY made similar errors in a cybersecurity report last month.
The fact that major consultancy firms are making such mistakes is far from harmless, says research group GPTZero, which scrutinized the report. KPMG and Deloitte are viewed as credible sources. The KPMG report, for example, had already been picked up by a Czech newspaper. Consultancy firms position themselves as the ideal partners to guide companies in responsible AI adoption, so we should expect them to be rigorous with themselves first and foremost.
