OpenAI's latest image-generating model, 4o, has improved significantly, capable of creating realistic text in images, which increases the risk of generating fraudulent documents. Users have leveraged this capability to create fake receipts and prescriptions, demonstrating the potential for expense fraud and sale of controlled substances. This advancement raises concerns over the challenges of detecting such faked documents, as current guardrails can be bypassed. Historically, earlier models struggled with text accuracy, but 4o's enhancements suggest that verification systems may soon need to evolve rapidly to keep up.
"You can use 4o to generate fake receipts. There are too many real-world verification flows that rely on 'real images' as proof. That era is over."
"Another user managed to edit the image further by adding a realistic filter and food stains - the perfect way to commit expense fraud, if you were so inclined."
"The development highlights how far AI-powered image generators have come. Previous models infamously struggled with recreating letters, often resulting in garbled shapes and unintentionally hilarious phrases."
"Whether our ability to detect this storm of faked documents remains to be seen. Guardrails like appended metadata are easily overcome."
Collection
[
|
...
]