A wave of concern has hit the Indian digital security landscape after social media users began sharing eerily realistic Aadhaar and PAN cards generated using GPT-4o, the latest AI model launched by OpenAI.
The model’s new image generation feature has already produced over 700 million images, according to Financial Express, sparking fresh fears around identity fraud.
GPT-4o allows users to create images directly within ChatGPT using simple prompts — an architectural upgrade over previous versions like DALL·E. However, this leap in generative ability appears to come at a cost. In its official system card, OpenAI acknowledges the tool carries “higher risks of misuse,” especially when it comes to documents that mimic official ID formats.
Several users have already posted images of fake Aadhaar cards featuring their own photos or, in one notable case, a card with Elon Musk’s name and UID — a viral post that quickly spread across social platforms.
Though some image elements, like facial features, may appear off, the overall design, including fonts, layout, and structure, has been strikingly accurate, Financial Express reports.
The issue extends to Permanent Account Number (PAN) cards as well. Experts warn that while Aadhaar enjoys a strong backend verification system, PAN and driving licences lack equivalent rigour, making them more susceptible to forgery.
Although OpenAI has implemented some restrictions, such as banning realistic depictions of children, public figures, and explicit content, it currently does not prohibit generating images of ID cards. With little regulation in place, experts are calling for urgent updates to AI governance before digital identity fraud becomes a full-blown crisis.
