Editing clues without an authenticity verdict

Photo Editing Detection

Check a still photo for readable software clues and separate visual model signals. Learn why ordinary retouching, generative editing, and a misleading caption are different questions—and why a low score does not prove a photo is untouched.

Still images only. One-time purchase, no subscription or daily free allowance. Save your private recovery code before paying.

$4.99 for 20 image checks

One-time purchase. No subscription. No account required.

Loading your balance…

Already purchased? Restore your checks

Drop a photo here

JPG, PNG, or WebP. Each completed model analysis uses one image check.

Edited does not mean generated, and neither means deceptive

Photo editing detection begins by separating different changes. A crop alters composition, exposure adjustment changes appearance, compositing combines sources, and generative fill may replace or extend image content. These operations can occur in the same editing application. Knowing that an editor handled a file does not identify which operation happened or whether the resulting photo misrepresents a claim. Ask what changed and why that change would matter in context.

The upload area on this page runs FauxGuard's existing image check. It reads available software, camera, and generator text and requests AI-generation and face-manipulation scores. It is not an all-purpose detector for every manual photo edit. A low model score can accompany extensive conventional retouching. A high score can justify additional review without establishing that a particular object was added, a background was replaced, or a person intended to deceive anyone.

What an editing software tag can and cannot tell you

A readable software field may name an editor or an export workflow. That is an observation about file text, not authenticated evidence of the action taken. An application can write its name during an ordinary resize, and the same name may appear after a more significant composite. Tags can also be copied, changed, or removed. They do not identify the editor, date every operation, or establish which parts of the image came from a camera.

FauxGuard inspects the processed upload rather than reconstructing the source file. Browser resizing and conversion can remove metadata before inspection, so keep an original copy for any independent metadata review. A missing software tag does not rule out editing; a camera field does not rule out later manipulation. A useful report should state these limits beside the observations rather than translating the presence or absence of a tag into a final authenticity label.

Generative fill is not the same as a whole-image AI score

A small generative change can occupy only a limited part of an otherwise camera-origin scene. The global AI-generation score may not detect it, and a high score cannot identify its boundaries. Conventional edits can also change a large area without creating the model patterns associated with generation. Use the score to guide a question about the file, not to claim a complete map of all edits or a count of altered pixels.

The face-manipulation score is reported separately because a modified likeness is another kind of signal. Neither number is a general measure of how much Photoshop work occurred. Confidence bands reflect score ranges, not validated edit-detection accuracy. FauxGuard does not provide an error-level analysis map, local tamper heat map, layer recovery, or original-image reconstruction. If you need to compare a specific region, seek an earlier version or a qualified review of the original materials.

Compare the image with the claim, not just with an ideal camera file

For a product listing, ask whether the photograph fairly represents the item being sold. A clean backdrop may be ordinary presentation work; a removed defect may be relevant to a dispute. The same editor can perform either change. For a portrait, retouching may change appearance without proving identity substitution. For an event image, cropping may alter the impression while leaving every remaining pixel camera-origin. The model cannot resolve those contextual judgments for you.

Preserve the received file and the accompanying message, then request an original or compare independently sourced versions when appropriate. Record differences in framing, objects, reflections, and lighting without assuming every difference is evidence of a generative tool. Repeated compression, filters, and export settings can also affect appearance. A disciplined comparison states what is observable and what remains a hypothesis, giving the other party an opportunity to provide context instead of relying on a single score.

  • Ordinary retouching: may be legitimate and may receive a low AI score.
  • Compositing: requires source comparison; software text cannot locate changes.
  • Generative editing: may produce model signals but can also be missed.
  • Misleading caption: can occur without any pixel editing at all.

Read the result before choosing a follow-up check

Separate the model inference from the file clues. A generator name is unverified text rather than confirmed attribution. C2PA-related text is not an authenticated edit history: FauxGuard does not validate its signature, and SynthID watermark verification is not supported. If model analysis is unavailable, the partial metadata result is not evidence that the photo passed an editing test. No image check is used for that fallback, and the notice explains the limit.

After scanning, decide which question remains unanswered. An earlier copy can help investigate a changed background; source research can clarify a misleading caption; a face-manipulation review can explain a portrait signal. Do not combine those observations into an unsupported legal or factual conclusion. For consequential disputes, preserve the originals and involve a qualified reviewer. The linked guides help you choose a proportionate next step without implying that this checker can reconstruct every change to a photograph.

Detector

Check an image with FauxGuard

Upload a JPG, PNG, or WebP image and review separate model scores, score-based confidence, readable file clues, and reasons. No account is required. A score is evidence to review, not proof of origin.

Related image checks

Frequently asked questions

Can FauxGuard prove a photograph has never been edited?

No. Missing software tags and low AI-generation or face-manipulation scores do not rule out manual edits. The tool does not reconstruct layers or establish an editing timeline. Compare original materials and source context if an untouched-file claim matters.

Will the checker highlight the exact edited area?

No local edit heat map or changed-region overlay is returned by this workflow. File clues and image-level model scores cannot reliably locate a replaced object or background. Request an earlier version for comparison or seek a qualified examiner for a more specific investigation.

Do I need an account to check an image?

No account or login is required. Your image-check balance is accessed through a private recovery code and a browser cookie. Save the code before paying and keep it private: anyone with it can spend your checks. You can restore the balance from the detector or pricing page on another device.

How much does an image check pack cost?

FauxGuard sells 20 image checks for $4.99 USD as a one-time purchase. There is no subscription or daily free allowance. One completed model analysis uses one check. Failed model requests and metadata-only fallback results do not use a check. Review pricing and the terms before paying.

How should I read the detection result?

Read the AI-generation score and the separate face-manipulation score alongside the reasons and file clues. Confidence labels describe score ranges, not a validated accuracy rate. A high score is a reason for further review; a low score does not certify authenticity. An unavailable score is not a low score.

What are the limits of this image check?

False positives and false negatives are possible, especially after resizing, compression, retouching, or changes in image generators. FauxGuard does not verify C2PA signatures or SynthID watermarks, identify a person, or provide a local edit heat map. It accepts JPG, PNG, and WebP still images, not video or audio. Results are informational, not forensic or legal findings.