About Fake Image Detector

Our mission is to give everyone, not only forensic specialists, a free and fast way to check whether an image is authentic, edited or AI-generated before they believe it or share it.

Since 2020
Online and free to use the whole time.
Four layers
Every scan runs four separate forensic checks.
No account
Nothing to sign up for and nothing to install.
Nothing stored
Images are analysed in memory, then discarded.

What the tool does

You upload an image and get a verdict within seconds, with the evidence from each layer behind it.

  1. AI detection

    A neural network estimates whether the image came from a camera or was made on a computer.

  2. Metadata analysis

    Reads the EXIF fields that name the camera and the software that saved the file.

  3. Error Level Analysis

    Recompresses the image and maps the areas that respond differently.

  4. Watermarks and credentials

    Checks for invisible AI watermarks and C2PA Content Credentials.

Our experience

Fake Image Detector has been online since 2020. It began as a focused ELA and metadata scanner and has grown, release by release, into the four-layer pipeline described above. Running a public detection tool for years has taught us where automated forensics shines and where it struggles, and that experience shapes both the analysis engine and the plain-language reports it produces.

How we build and maintain the detector

Our detector is trained on a large set of camera photographs and computer-made images that we curate and expand over time, and it is tested on images from AI generators it never saw during training. As new generators appear, we retrain it and adjust the pipeline. Detection quality is a moving target, so maintenance is continuous rather than occasional.

Honest limitations

No automated detector is perfect, and we would rather tell you that plainly than pretend otherwise. Sophisticated edits can evade individual techniques: metadata can be stripped, AI artifacts can be smoothed away, and watermarks only exist when a generator embeds them. Treat our results as strong signals to inform your own judgement, not as conclusive proof, and see our disclaimer for details on how to interpret them.

Editorial standards

Alongside the tool we publish a blog on image forensics, deepfakes, and digital authenticity. Articles are researched against how detection actually works in practice, written to be accurate without hype, and updated when techniques or standards change. If you spot an error in anything we publish, we want to hear about it.

Privacy by design

Every image you scan is analysed entirely in memory. Nothing is written to disk, stored on our servers, or used to train our models. The moment your result is generated, the image is discarded. There are no accounts, no sign-ups, and no personal information required to use the tool.

Read the full details in our privacy policy.

How the site is funded

Fake Image Detector is free to use and is funded in two ways: advertising served through Google AdSense, and voluntary donations from users through Ko-fi. Neither ads nor donations influence analysis results, and we do not sell user data. Keeping the tool free matters to us: image verification should be available to everyone, not just newsrooms with forensic budgets. If the tool has helped you, supporting it on Ko-fi helps cover the cost of running and improving it.

Support us on Ko-fi (opens in a new tab)

Contact us

Questions, corrections, or feedback are always welcome.

Have a photo you are not sure about?

Run it through all four forensic checks. It is free and needs no account.

Scan an image