KPMG Fabricated AI Case Studies in a Report Selling Clients on AI
KPMG is one of the world’s biggest consulting firms. A consulting firm is a company that other companies pay for expert advice. KPMG got caught using fake AI case studies in one of its reports. “Fabricated” means made up or false. A case study is a real-life example used to show that something works.
The report was meant to get clients to start using AI. But it named real companies and said they did things they never did. This is a clear warning. AI tools can make up facts. This can happen even in work from big, trusted brands.
The report was called “Redefining excellence in the age of agentic AI.” Agentic AI means AI that can act on its own to finish a task. It does not just answer one question. KPMG sells AI advice as a service. So a report full of false AI examples looks very bad for them.
What went wrong
The report named real organisations as proof that AI works. These were the Swiss bank UBS, the UK’s National Health Service (NHS), Swiss Federal Railways, and Transport for London. But all of these organisations said the claims were wrong. In simple words, they said the stories about them were not true.
The fake case studies were found by a company called GPTZero. GPTZero builds tools that spot text written by AI. Then the Financial Times, a big newspaper, checked the errors on its own. It found the same problems. The report had a date of June 14, 2026.
Key facts
| Detail | As reported |
|---|---|
| Report title | “Redefining excellence in the age of agentic AI” |
| Publisher | KPMG |
| Organisations falsely cited | UBS, UK NHS, Swiss Federal Railways, Transport for London |
| Who found it | GPTZero |
| Who verified it | Financial Times |
| Publication date | June 14, 2026 |
How the mistakes happened
GPTZero looked into the report. It found that the sourcing was sloppy. Sourcing means showing where a fact comes from. The report used AI search tools in a careless way.
It copied real sources in a loose way. Some had no web links. Some had the wrong authors. And some notes pointed to no real source at all.
GPTZero gave this habit a name: “vibe citing.” It means picking sources based on a rough feeling, not real checking. A citation is the note that tells readers where a fact came from. When citations are fake or wrong, readers cannot trust the fact.
The bigger danger: secondary hallucinations
The CEO of GPTZero, Edward Tian, warned about “secondary hallucinations.” A hallucination is when an AI says something false as if it were true. A secondary hallucination is what comes next.
Here is how it works. Bad reports from big firms get reused by AI systems and by people. The false facts then spread further and further. So one mistake can grow into many.
After the problem came out, KPMG took the report off several of its websites.
Why it matters (especially for India and founders)
Many Indian founders and teams are rushing to use AI. They use it for research, pitch decks, and reports. A pitch deck is a short slide show used to explain a business idea. This story shows the risk.
AI can save time. But it can also slip in fake facts without you noticing. If a top global firm like KPMG can get caught, a small startup can too.
The fix is simple but very important. Check every fact and link before you publish. Treat AI as a fast first draft, not a final source of truth. If you sell AI services, trust is your product. One fake case study can wipe out years of trust.
FAQ
What did KPMG do wrong?
KPMG put out a report with fake AI case studies. It said real organisations, like UBS and the UK’s NHS, did things they later said were not true.
Who discovered the fabricated case studies?
GPTZero found them. It is a firm that detects text written by AI. The Financial Times then checked the errors on its own and confirmed them.
What is “vibe citing”?
It is a term GPTZero used. It means picking sources based on a rough feeling, not real checking. It led to citations with no real source behind them.
What did KPMG do after the report was exposed?
KPMG took the report off several of its websites.
The takeaway
This case is a warning for the whole AI era. The tools are powerful. But they still need humans to check them.
A report made to sell AI ended up showing why careful, checked work still matters. Always confirm facts and sources first. Do this before you trust, or publish, anything written with AI.
Source: The Decoder
