Fake Aadhaar cards spark concern as ChatGPT’s image tool hits 700 million creations

Since OpenAI unlocked GPT-4o’s native image generation last week, users have created over 700 million images, with Studio Ghibli-inspired artwork leading the viral trend. But the creative enthusiasm is now being met with a growing concern: fake Aadhaar cards.

Social media platforms have begun seeing AI-generated Aadhaar-like images, allegedly produced using ChatGPT’s new image generator. Some posts even featured photorealistic mock-ups of India’s unique ID card, showing fabricated names and portraits ranging from ordinary users to public figures like Sam Altman and Elon Musk with QR codes and ID numbers attached.

We tested it. ChatGPT could create a layout resembling an Aadhaar card, complete with the red branding bar and government motifs. While the faces lacked realism due to current limitations, the resemblance to the original document was striking enough to raise alarm.

This scenario underlines a long-standing concern: the risk of AI misuse when powerful generative tools are released without sufficient guardrails.

Unlike earlier versions that used external tools like DALL·E, GPT-4o generates images natively from within ChatGPT. This enables it to follow complex natural language instructions and produce more accurate, nuanced, and stylistically diverse visuals, but also increases the likelihood of misuse.

The company maintains strict restrictions around creating photorealistic images of minors, violent or abusive content, and explicit imagery. However, synthetic identities and deepfake-style impersonations still fall into a grey area, especially when it comes to fake documentation or ID cards, which have real-world implications for fraud and misinformation.

The rise in such examples points to the urgency of enforcing ethical AI frameworks, particularly in regions like India, where digital identity is integral to public services and financial inclusion. As generative AI tools become more powerful and accessible, authorities and platforms may need to rethink digital verification mechanisms to counter potential manipulation.

For now, ChatGPT’s internal safety checks block certain high-risk prompts, but workarounds and partial fabrications remain a challenge. In an era where realism is only a prompt away, it’s clear that AI governance will need to evolve just as quickly as the tech itself.