A checker for one image, not a certificate of authenticity
Use this AI-generated image checker when you have a specific picture and want an additional signal about its origin. You can choose a supported image in the upload area without creating an account. The analysis combines readable file clues with separate visual model scores. This page focuses on interpreting those outputs so you can decide whether another source check, an original file, or human review would be useful.
The question is narrower than whether a picture is truthful. A camera photo can be staged or paired with a misleading caption. A generated illustration can be honestly labeled. FauxGuard does not decide either case from pixels alone. It also does not turn ordinary editing software metadata into an AI-generation probability. Keep the file's origin, its editing history, and the claim attached to it as separate questions throughout your review.
Prepare a useful upload without discarding your source
Choose a JPG, PNG, or WebP that contains the actual image you want to evaluate. Preserve the received file elsewhere, including its source link when available. If you only have a small thumbnail or a screenshot of an entire social feed, ask for a better copy before drawing strong conclusions. Interface text and borders are not evidence about how the underlying photo was made, and compression can obscure useful visual patterns.
FauxGuard's browser processing reduces large images to a maximum edge of 2048 pixels and may convert them to WebP. That improves upload practicality but can remove metadata before the server reads it. A missing camera or editor field in the report therefore cannot establish that the original lacked such information. Review original metadata independently when it matters; this checker does not preserve a chain of custody or provide an authenticated record of the source file.
How to interpret high, mixed, and low AI-generation scores
The AI-generation model score runs from 0 to 100. FauxGuard labels scores of 90 or above as likely AI-generated, scores above 10 and below 90 as mixed or inconclusive, and scores of 10 or below as low AI-generation signal. These interface labels organize a review; they do not establish the true origin of every image in a score range. False positives and false negatives remain possible.
The confidence band is also derived from the score range, not a separate accuracy test for your upload. High score-based confidence can accompany either end of the range, including a very low AI-generation score. Do not confuse it with an independent statement that a file is authentic. Read the reasons, source context, and the separate face-manipulation score before deciding whether a result supports the particular question you are investigating.
- High score: investigate the source rather than making an accusation.
- Middle score: keep the origin unresolved and seek additional evidence.
- Low score: the model found weak AI signals, not proof of authenticity.
- No score: the model did not provide a usable answer; read the notice.
Keep file clues and face-manipulation evidence separate
A face-manipulation score addresses a different model signal from whole-image AI generation. A photograph with an altered face may not look like a fully generated scene. A synthetic image may have no relevant face at all. Do not average the scores into a new authenticity rating or treat a low face score as confirmation that the depicted person is genuine. This tool does not perform identity verification or identify who made an edit.
Software names, generator names, and camera fields are readable clues rather than authenticated facts. They can be absent, removed, copied, or changed. FauxGuard may report C2PA-related text, but it does not verify the credential signature, and it does not check SynthID watermarks. A generator-family clue describes a name found in file text, not a verified model attribution. None of these fields locates the exact region of an edit.
When a partial result or another version changes the answer
If the visual model fails or returns incomplete scores, FauxGuard displays a partial result and does not use an image check. Any remaining metadata is still limited evidence; absence of a score is not a passing result. Read the notice before retrying. Service capacity can temporarily delay analysis, and buying a pack does not guarantee an immediate model response or a particular conclusion.
Two versions of an image can produce different scores because their processing differs. Record which file you checked instead of selecting the result you prefer. Repeatedly scanning the same file does not supply independent evidence about origin. When the model result leaves a question unanswered, source research is often more useful than another score. The linked guide explains how to preserve your observations, compare earlier versions, and make an appropriately limited summary.
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 92/100 score mean a 92% chance the image is AI?
No. It is the model score after normalization to a 0–100 scale, not a calibrated probability for your file. FauxGuard labels that range likely AI-generated, but the reasons and source context still matter. The confidence label is score-based and is not a separately validated accuracy claim.
Will checking the same picture repeatedly make the answer more reliable?
Not by itself. Repeated responses from the same model are not independent source evidence. Re-encoding or cropping can change the input and its score. Preserve the original, note which version was tested, and seek a source or independent review rather than selecting a preferred number.
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.