Check any image for AI in seconds
Our AI image detector looks at a picture two ways. First it reads the file for hidden labels that AI tools leave behind — Content Credentials (C2PA), IPTC “trainedAlgorithmicMedia” tags, and the generation settings Stable Diffusion and ComfyUI save into PNG files. Then it runs the pixels through several open-source detectors trained on millions of real and AI images.
Works with JPG, PNG and WebP up to 15 MB. On iPhone, photos are converted automatically when you pick them.
What the result means
| Verdict | What it means |
|---|---|
| AI-generated (confirmed) | The file carries a label from an AI tool. Essentially certain. |
| Likely AI-generated | The detectors strongly agree the image was made by AI. |
| Possibly AI-generated | Some AI traits — could be AI or heavily AI-edited. |
| Hard to tell | Mixed signals. Try to find the original source. |
| Probably a real photo | Mostly natural signals with a few uncertain areas. |
| Likely real, not AI | No meaningful AI signals found. |
Tips for the most accurate result
- Use the original file whenever you can. Social apps strip labels and compress images, which makes detection harder.
- Crop screenshots to just the image — remove app buttons, captions and borders.
- Check the biggest version of the image you can find, not a thumbnail.
- Check more than once if it is a famous image: different copies online may carry different evidence.
Images we can detect
ChatGPT and DALL·E 3, Google Gemini (Nano Banana) and Imagen, Midjourney v6/v7, Stable Diffusion 1.5/SDXL/SD3, Flux, Adobe Firefly, Microsoft Copilot Designer, Meta AI, Grok, Ideogram, Leonardo and more. AI-edited photos (generative fill, object removal) are flagged when the editing app records it in the file.
Frequently asked questions
Can this detect AI images from ChatGPT?
Yes. ChatGPT images that still have their Content Credentials are confirmed instantly. Screenshots and re-saved copies are checked by the pixel detectors.
Can I check a screenshot or an image I found on social media?
Yes. Screenshots and re-saved social media images lose their file labels, so the result relies on the pixel detectors alone. Crop to just the picture you want to check (no app buttons or captions) for the best result.
How accurate is it?
When a file still carries Content Credentials or other AI labels (common for ChatGPT, Gemini, Firefly and Stable Diffusion files) the answer is essentially certain. Otherwise we combine several open-source detectors; they are right most of the time on unedited AI images but can be fooled by screenshots, heavy compression, filters and brand-new generators. That is why every result shows a confidence level and the evidence behind it — treat it as strong evidence, not proof.
Why did a real photo come back as AI (or the other way round)?
Heavy beauty filters, HDR or “AI enhance” modes on phones, upscalers and strong compression can all look synthetic to a detector. Very new or heavily edited AI images can slip through. Look at the confidence level and the “Why” list, and use the 👍 / 👎 buttons — your feedback helps us tune the detectors.
Do you keep my photos, videos or text?
No. Files are analyzed in memory and thrown away as soon as your result is ready. Videos never leave your device — your browser pulls out a few frames and only those are checked. Nothing is published unless you tap Share, and then only a small preview is saved.
Is IsItMadeByAI really free?
Yes. Every scan is free, there is no sign-up, no credit card and no watermark on your results. The site runs on open-source detection models, which is how we keep it free.