ChatGPT’s new image generator is really good at faking receipts

This month, ChatGPT has unveiled a new images generator as part of its 4O model, which is much better to create a text inside the images.

People are already used to create a fake restaurant receipts, which may add another tool to the already stretch tools of AI Deepfakes used by fraudsters.

The abundant social media poster and VC Deedy Das have been published on x a picture of a fake receipt of the San Francisco Steakouse that he says was created with 4O.

Others managed to repeat similar results, including results with food or drink to make them look more authentic:

The most realistic example of Techcrunch was actually from France, where the LinkedIn user has published a generated receipt of artificial intelligence for a local restaurant series:

Techcrunch 4O was tested and was also able to create a fake receipt for Applebee’s in San Francisco:

But our attempt had dead gifts that were fake. For one, the total is used as a comma instead of a while. For another, do not add mathematics. LLMS is still struggling to do basic mathematics, so this is not particularly surprising.

But it will not be difficult for fraud fraudulent to fix some numbers quickly using photo editing software, or, perhaps, more accurate claims.

It is clear that it is really easy to generate fake receipts representing great chances of fraud. It will not be difficult to imagine this type of technology used by bad actors to get a “compensation” for fake expenses completely.

Openai Taya Chrostianson techcrunch spokeswoman told a spokeswoman that all her photos contain definition data indicating that she made it ChatGPT. Christianson added that “Openai” takes action when users violate their own use policies and that he “always learns” from real use and comments.

Techcrunch then asked why Chatgpt allowed people to create fake receipts in the first place, and whether this is in line with Openai’s use policies (which prohibits fraud.)

Christianson replied that “Openai’s goal is to give users as much creative freedom as possible” and that fake artificial intelligence receipts can be used in non -Fraud situations such as “teaching people about reading and financial writing” as well as creating original art and product ads.