ChatGPT Can Fake A Receipt But Can’t Spot Its Own Work
ChatGPT can forge receipts so convincingly that even ChatGPT can't detect them. Here's why that matters for your fraud detection strategy.
Why it matters
New research reveals a critical blind spot in AI safety and fraud detection: LLMs can generate synthetic documents that defeat both human review and AI-powered verification tools, raising serious questions about authentication systems in a world of synthetic content.
The key facts
6 to knowChatGPT can generate forged receipts indistinguishable from authentic ones
Forensic tools cannot differentiate fake receipts from real ones
ChatGPT itself cannot identify its own synthetically generated receipts
Humans unable to spot the forgeries in blind tests
Published May 2026 in Forbes by Lars Daniel
Highlights critical gap in AI safety and document authentication
Go to the source
Forbes Innovationforbes.com
Publisher excerpt: New research shows ChatGPT can forge receipt numbers so convincingly that humans, forensic tools, and even ChatGPT itself cannot tell the fakes from originals.