How to check if an image was made with AI
- Drop, paste or select an image: JPG, PNG, WebP, HEIC or AVIF.
- Read the verdict at the top: Made with AI, Edited with AI, Possible AI signs or No AI labels found.
- Look at the details: Content Credentials, the AI prompt if one is stored, and the camera, date and location data.
- If the file contains location or personal data, use the Image Metadata Remover before sharing it.
What this checker looks for
Many AI image generators now label what they create, but the labels are hidden inside the file where you can’t see them. This tool reads four kinds of evidence:
- Content Credentials (C2PA). A signed record that names the app and lists what it did. Images from ChatGPT, DALL·E, Adobe Firefly and Microsoft Designer include one. If its actions say the image was created by a “trained algorithmic” model, it was generated by AI.
- IPTC digital source type. A standard photo tag with values such as “trainedAlgorithmicMedia”. Google, Meta and Adobe use it for AI images.
- Generation settings. Stable Diffusion tools save the prompt, seed and model in PNG text chunks. When we find them, we show the prompt.
- Software tags. Names like Midjourney, Firefly or DALL·E in the Software or Creator Tool fields. These are treated as hints, because they are easy to change.
Why “No AI labels found” doesn’t mean the image is real
Labels only help while they stay in the file. Taking a screenshot, re-saving the image in most editors, or uploading it to WhatsApp, Instagram, Facebook or X usually strips all hidden data, including AI labels. Some generators, including many open-source models, never add labels at all. So a missing label tells you nothing either way. Treat this tool as a quick first check, and for important decisions also look for the original source, reverse image search the picture, and check for visual errors like garbled text or odd hands.
Content Credentials explained
The C2PA standard was created by Adobe, Microsoft, Google, OpenAI, the BBC and others to show where digital content comes from. A Content Credential is a small manifest stored in the file that records the app, the actions taken (created, edited, cropped, placed an element) and, for AI, the type of source. Each manifest is signed with a certificate that names the company. This tool reads the manifest and the certificate name right in your browser. It does not cryptographically verify the signature, so treat it like reading a label, not like checking a passport.
Your photos’ other hidden data
While checking for AI labels, the tool also shows the rest of the file’s metadata: the phone or camera model, lens and settings, the date and time it was taken, the editing software and, most importantly, GPS coordinates. Photos taken with location services on can reveal your home address to anyone you send the original file to. If you see a Location card, remove the data before posting the photo on a forum, marketplace or website.
Frequently asked questions
Can this tool tell for certain if an image is AI-generated?
It can confirm it when the file carries an AI label, such as Content Credentials or an IPTC "trained algorithmic media" tag. When no label is found, it cannot prove the image is real, because labels are easy to lose: a screenshot or a re-upload to most social apps removes them.
Is my image uploaded anywhere?
No. The file is read by your own browser and never leaves your device. You can check this in your browser’s developer tools: no image data is sent over the network.
Which AI tools add labels to their images?
OpenAI (ChatGPT and DALL·E), Adobe Firefly, Microsoft Designer and Bing Image Creator add C2PA Content Credentials. Google adds IPTC labels to images from Gemini and Imagen, and Meta labels images from Meta AI. Stable Diffusion tools like Automatic1111 and ComfyUI save the prompt and settings in PNG files.
Can it detect Google’s SynthID watermark?
No. SynthID is invisible and hidden in the pixels themselves; it can only be read with Google’s own detector. This checker reads the labels stored in the file, which are separate from SynthID.
What are Content Credentials?
Content Credentials (the C2PA standard) are a signed record stored inside a file that says which app created or edited it and what was done, for example "created with a generative AI model" or "cropped in Photoshop".
Why does it say “Possible AI signs”?
The file mentions AI software, for example in its Software tag, but carries no formal AI label. That is a strong hint but not proof, because anyone can edit a Software tag.