ChatGPT generates images using DALL-E 3 — and TruthLens can detect them. Upload any suspicious image to find out if it was made by ChatGPT, Sora, or any other OpenAI tool.
When you ask ChatGPT to "generate an image" or "draw me...", it uses OpenAI's DALL-E 3 model under the hood. The resulting images have specific visual characteristics — a distinctive smoothness, particular color palette tendencies, and a tendency toward hyperrealistic but slightly uncanny subjects.
TruthLens has been trained to recognize these patterns across thousands of ChatGPT-generated images, and can identify them with 90%+ confidence even after JPEG compression or social media resizing.