• Home
  • Blog
  • AI News
  • ChatGPT New Image Generator Can Create Fake Receipts, Here’s Why That’s a Problem

ChatGPT New Image Generator Can Create Fake Receipts, Here’s Why That’s a Problem

Published:April 1, 2025

Reading Time: 2 minutes

OpenAI just launched an upgraded image generator in its ChatGPT-4o model and it’s great at adding text to images. But there’s a problem. People are already using it to make fake restaurant receipts. These look shockingly real.

AI-Generated Receipts Flood Social Media

Tech junkies noticed fast. Venture capitalist, Deedy Das, posted a fake receipt from a San Francisco steakhouse. He said he made it with ChatGPT-4o, and the image spread quickly, prompting others to join in. Some even added food stains and crumpled effects. One LinkedIn user in France even shared a crinkled AI-generated receipt from a local restaurant.

TechCrunch ran its own test. They generated a fake Applebee’s receipt in San Francisco but it wasn’t perfect. The total used a comma instead of a period. Even worse, the math didn’t add up. That is not surprising given LLMs still struggle with numbers.

A Growing Concern for Fraud Prevention

These mistakes may not stop fraudsters. Fixing a few numbers with editing software is easy. A polished fake could fool many.

Here’s where this could lead:

  • Expense fraud: Employees could submit fake receipts for reimbursement.
  • Tax fraud: People might create receipts to boost deductions.
  • Refund scams: Scammers could request refunds for meals they never bought.

OpenAI’s Stance on the Issue

TechCrunch asked OpenAI about this. Spokesperson Taya Christianson confirmed all AI-generated images include metadata. She said OpenAI watches how people use the tool and takes action when needed.

But why allow fake receipt generation at all? OpenAI’s policy bans fraud. Christianson explained that OpenAI values creative freedom. She said fake receipts could be useful for education, art, or ads.

AI’s Role in Financial Integrity

AI-generated deepfakes are growing, and businesses and regulators need to keep up. Fraud prevention teams are increasingly encumbered with finding new ways to detect AI forgeries.

Possible solutions:

  • Using AI detection tools to spot fake text and images.
  • Training employees and consumers to recognize AI fraud.
  • Strengthening verification for refunds and reimbursements.

Lolade

Contributor & AI Expert