About FreeAIDetector

What it does

FreeAIDetector.org checks whether content was made by AI. Each detector looks at the evidence that suits its kind of content:

• AI Image Detector. Checks photos, screenshots and art for signs of AI generation, using file metadata, pixel forensics and a vision model.
• AI Video Detector. A first look at a video clip: frames are taken in your browser and scored one by one by the same vision model as the image detector. In beta.

Every result comes with the reasons behind it, and the detectors follow the same rules: they say “uncertain” instead of forcing a guess, and none of them needs an account.

How it works

Images

• File metadata. Your browser reads provenance clues inside the file: C2PA Content Credentials, the IPTC DigitalSourceType field, EXIF software and camera data, and generation parameters left by tools such as Stable Diffusion. A clear AI marker settles the result on the spot.
• Pixel forensics. The browser also measures grain, sharp detail and flat areas. That tells a screenshot from a photo and adds context, but texture alone can't prove an image is AI.
• Vision model. When the file isn't conclusive, the image goes to our server, where a classifier trained on real and generated images scores it.

Video (beta)

Your browser takes four to eight evenly spaced frames from the video, and each frame is scored by the same vision model that checks images. The scores are combined into one result that leans toward the most suspicious frames. The video file is never uploaded: only the frames are sent, analyzed in memory and discarded. It is a first look. It doesn't read the labels inside a video file, it can't see motion or hear sound, and it hasn't yet been tested on a set of labeled videos, so we publish no accuracy figure for it.

What it can't do

No detector is infallible. Metadata can be stripped or faked, and compression, editing and illustration styles can mislead both pixel analysis and models. Video checks look at only a few frames and use a model built for still images, so they can miss edits, face swaps and anything that happens between the frames. Treat a result as one piece of evidence, and weigh it against the source and context before drawing conclusions about anyone.

The guides explain how AI detection works and where it fails, with sources.

Why it's free

Checking whether something is real shouldn't require an account, a subscription or a credit pack. The detectors are free to use, with fair-use limits that keep them available for everyone.

Get in touch

Found a wrong result or have a suggestion? Contact us. Reports of wrong results show us where the detectors fail.