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

Tech Crunch - Mar 31st, 2025

ChatGPT has launched a new image generator as part of its 4o model, which significantly improves the generation of text within images. This development has led to concerns about the ease with which fake restaurant receipts can be created, potentially aiding fraudsters. Venture capitalist Deedy Das demonstrated the tool's capabilities by posting a fabricated receipt of a San Francisco steakhouse on social media platform X. Others have replicated these results, even adding realistic touches like stains to enhance authenticity. TechCrunch also tested 4o, successfully generating a fake receipt with minor flaws, underscoring the potential for misuse.

The introduction of this advanced image generation tool by OpenAI raises significant issues regarding fraud and the verification processes that rely on image authenticity. Although OpenAI embeds metadata in images to indicate their AI origin and enforces usage policies against fraud, the technology's potential for abuse is evident. OpenAI's spokesperson, Taya Christianson, emphasized the company's commitment to user creativity and highlighted legitimate applications, such as financial literacy education. However, the ease with which fake receipts can be produced presents a challenge for industries reliant on visual documentation, necessitating enhanced verification measures to counter potential fraudulent activities.

Social Media Fraud Open Ai Tech Crunch Ai Image Generation Chat Gpt 4o Deedy Das Fake Receipts Financial Literacy

Story submitted by Fairstory

RATING

7.6

Fair Story

Consider it well-founded

The article provides a well-rounded overview of ChatGPT's new image generator, highlighting both its technological advancements and the ethical concerns it raises. It effectively uses credible sources, such as OpenAI and TechCrunch, to support its claims, though it could benefit from additional expert perspectives. The story is timely and relevant, addressing issues of public interest and potential societal impact. While the article is clear and engaging, it could enhance its engagement and impact by incorporating more interactive elements and exploring regulatory implications more deeply. Overall, the article succeeds in informing readers about a cutting-edge development in AI technology while prompting consideration of its broader implications.

Accuracy

The story accurately reports on the introduction of ChatGPT's new image generator, 4o, and its enhanced capabilities in generating text within images. This is supported by OpenAI's official announcements. The article correctly identifies the potential for misuse in creating fake receipts, which TechCrunch and other sources have verified through practical examples. However, some claims, like the potential for non-fraudulent uses, are speculative and not extensively backed by evidence. The story presents a mostly accurate depiction of the technical limitations of the 4o model, such as its struggles with basic math, aligning with known limitations of LLMs.

Balance

The article presents a balanced view by discussing both the technological advancements and the ethical concerns associated with the new image generator. It includes perspectives from OpenAI, highlighting their stance on creative freedom and fraud prevention, as well as concerns from users and experts about potential misuse. However, it could benefit from more diverse viewpoints, such as those from cybersecurity experts or consumer protection agencies, to provide a fuller picture of the implications.

Clarity

The article is well-structured and uses clear language, making it accessible to a general audience. It logically presents the information, beginning with the introduction of the new technology, followed by potential applications and ethical concerns. The tone remains neutral, focusing on factual reporting rather than sensationalism. However, some technical details, such as the specific limitations of LLMs, could be explained more thoroughly for readers unfamiliar with AI technology.

Source quality

The story relies on credible sources, including statements from OpenAI and practical testing by TechCrunch, which adds to its reliability. The inclusion of a spokesperson from OpenAI provides authoritative insight into the company's policies and intentions. However, the story could be strengthened by citing additional independent experts or studies to corroborate the claims about the potential for fraud and the technical capabilities of the 4o model.

Transparency

The article is transparent in its reporting, clearly stating the sources of its information, such as the testing conducted by TechCrunch and statements from OpenAI. It explains the basis for its claims about the image generator's capabilities and potential for misuse. However, it could improve by providing more detailed explanations of the methodology used in the practical tests and any potential conflicts of interest from sources, such as the vested interests of TechCrunch or OpenAI.