Still portraits, not identity verification

Face Manipulation Checker

Upload a still photo and review the face-manipulation model score separately from AI generation. A suspicious result is a lead for source research, not proof about a person's identity or intent.

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.

A face check is not an identity check

A face manipulation checker can add a model signal when a portrait appears to have been altered. It cannot verify that the pictured person owns an account, identify the person, or establish consent. Those are different questions requiring different evidence. An authentic photo can be stolen and used for impersonation without changing its pixels. A fictional synthetic portrait can also be presented honestly as an illustration rather than as a real person's identity.

Use the upload area on this page for a supported still image. The same FauxGuard detector returns AI-generation and face-manipulation scores separately, alongside readable file clues and reasons. The face score is not a face-recognition match, an identity confidence score, or a conclusion about a particular person's actions. Keep that distinction visible when explaining a suspicious portrait to a moderator, colleague, or support team.

How a face swap differs from a fully generated portrait

A portrait can involve several workflows: generating a new scene, changing a face in a camera photo, blending a likeness, or applying ordinary retouching. These workflows can overlap, and one score cannot reliably reconstruct the steps. A low whole-image AI-generation score does not rule out a manipulated face. Similarly, a high AI-generation score does not tell you whether a real person's likeness was used as a reference.

Read the face-manipulation score as a separate model observation, not as a second vote on the same question. The values should not be added or averaged into a stronger-looking verdict. A global face score also does not identify which face or region was altered when several people appear. FauxGuard does not return a local manipulation heat map or recover the original face. A comparison image or a credible source may be more useful than another scan.

Retouching, filters, and compression can complicate the result

A real face can look unusual after smoothing, heavy noise reduction, lighting changes, or repeated compression. A model may respond strongly even when the edit was not a face swap. Conversely, a convincing manipulation may receive a weak signal. Keep false positives and false negatives in mind rather than interpreting a dramatic score as self-explanatory. The reasons are an aid to review, not a complete explanation of a model's internal decision.

Choose the least-processed image available and preserve the original separately. FauxGuard resizes large images and may re-encode them before upload; this can remove metadata and change the representation evaluated by the model. A screenshot, thumbnail, or heavily cropped portrait contains different evidence from a source file. If you compare versions, label which result belongs to which image instead of selecting whichever score confirms your initial suspicion.

File clues cannot authenticate a portrait's history

Camera fields can suggest a photographed source, while an editor name may suggest processing. Neither establishes an intact chain from the camera to the final portrait. Tags can be copied or altered, and a camera-origin image can later receive a face edit. Missing fields are also common after exporting or sharing. In this workflow, absent metadata is a limit on what can be observed, not positive evidence that the person or photo is synthetic.

A readable generator name is shown as an unverified clue rather than a confirmed model attribution. FauxGuard does not validate C2PA signatures or SynthID watermarks. Confidence labels describe score ranges rather than a validated accuracy rate for faces. If analysis is unavailable or incomplete, a partial notice is displayed without using a check. An unknown result does not verify the portrait, and retrying the same image does not supply independent evidence.

What to do after a suspicious profile-photo result

Preserve the post or profile URL, check account history, and look for earlier uses of the image. A reverse image search can help investigate reuse, while a detector addresses model signals. Neither replaces a proportionate identity-verification process. Avoid uploading intimate, private, or otherwise highly sensitive material simply to investigate a suspicion, and do not circulate a portrait together with an accusation based only on an automated result.

For a consequential decision, ask what evidence is missing and provide a human review path. A person affected by an allegation should have a way to supply context or an original. If the issue concerns a video, remember that this tool checks still frames only and cannot evaluate voice or movement. Use the linked guides to separate source research, image analysis, and identity claims before deciding what any score means in the particular situation.

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

Does a low face-manipulation score prove a profile is genuine?

No. It describes a model signal in the submitted image, not account ownership or identity. Someone can reuse an unmodified real photo. Check the account and source context separately, and do not treat either model score as proof that a person is genuine.

Can the checker show exactly which face was edited?

No. FauxGuard returns an image-level face-manipulation score, not a per-person identity match or local edit heat map. When several faces are present, the result does not establish which one may have been altered. Source comparison and qualified review are needed for more specific claims.

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.