
A dating or social profile
Check whether a profile picture shows signs of AI generation before you trust it. A result about the image cannot prove who a person is, so reverse image search and a live video call still matter.
Check a profile photoIs this image AI-generated, AI-edited or Photoshopped? Free online, no sign-up.
JPG, PNG, WebP, HEIC or AVIF up to 20 MB · We don’t store your image
An AI image detector looks for signs that a picture was generated or edited with AI. AIDetectorAll checks the file’s labels, content credentials and editing traces, then uses AI detection when needed. See what we found and what remains uncertain. One check a day is free, with no sign-up.
The free check uses up to one independent detector. A full report adds available detector readings, generator clues and a downloadable PDF. We show the evidence behind the result rather than treating a score as proof. Read the full method in our checking methodology.
We look for content credentials (C2PA), supported AI labels, generator settings, camera data and editing history. Signed records and editable metadata are reported separately. Neither verifies that the scene or its caption is true.
The free check can use one detector when file evidence is insufficient. A full report attempts two independent readings, including when labels are present. Scores can be wrong; unclear or conflicting results may be Inconclusive.
AI-generated, AI-edited, Editing traces found, No signs of AI, or Inconclusive. Results based on detector readings or unverified labels are qualified; none certifies an image as authentic.
The result includes a check summary with its findings, date and method version. A full report adds detailed findings, available generator clues, PDF download and report sharing.
The answer separates generation from editing and leaves room for uncertainty. A missing AI label does not show that an image is real. Screenshots, sharing and editing can remove labels.
The whole image was generated.
A real photo with parts changed by AI.
The file records editing software. This does not rule out AI.
Nothing we can find points to AI or editing. Not proof it is real.
The available evidence does not support a clear answer.
Each example shows an actual check and its main finding. These examples explain the results; they are not an accuracy benchmark.

Signed label from Google says it was made with AI.
Made by AIDetectorAll with Google Gemini from the CC0 beach photo Source / license

The file says it was edited with AI.
CC0 beach photo with the harness recoloured by Google Gemini, composited by AIDetectorAll Source / license

Saved by GIMP 3.2.8.
CC0 beach photo cropped and brightened in GIMP by AIDetectorAll Source / license

The detector sees no signs of AI.
Photo: Wikimedia Commons · CC0 Source / license

The detectors disagree.
Phone-size screenshot of a Google Gemini edit of the CC0 beach photo, made by AIDetectorAll Source / license
An AI photo detector has more to work with in the original file. Screenshots and social media copies often lose camera data, labels and image detail. They can still be checked, but the result may be less reliable. Here is what each kind of copy typically retains.
| Evidence | Original | Screenshot | Social copy |
|---|---|---|---|
| Content credentials (C2PA) | Often preserved | Usually removed | Usually removed |
| Camera and EXIF data | Usually preserved | Replaced or removed | Often removed |
| AI labels and generator settings | Preserved when present | Usually removed | Usually removed |
| Pixels for detector readings | Full available detail | Recompressed detail | Resized and recompressed |
These are common outcomes, not guarantees: platforms and export settings differ. Whenever you can, upload the original. In a messaging app, send the photo as a document or file instead of a picture, so it is not recompressed.
Checking a phone or camera photo? AI photo detector
You can check images from Midjourney, ChatGPT, Gemini (Nano Banana), Flux, Stable Diffusion, Seedream and Doubao. The result depends on the labels the file still carries and what the detector recognizes. We cannot reliably identify every generator, model or small AI edit.

Check whether a profile picture shows signs of AI generation before you trust it. A result about the image cannot prove who a person is, so reverse image search and a live video call still matter.
Check a profile photo
Look for editing software, saved history and AI edit labels in the file. Ordinary cropping and color correction are edits too; a trace does not by itself mean the content is deceptive.
Check for editing
Separate AI generation, AI edits and traditional edits instead of forcing every image into a real-or-fake answer. When the evidence is split or too thin, the honest result is Inconclusive.
Check a viral imageIt also works as an AI art detector for digital art, illustrations and anime-style images. Stylized images leave detectors less to work with, so expect Inconclusive more often.
Check for signs of AI generation and photo editing.
Find what a generator left inside the file.
Read camera details, editing history and credentials in plain language.
Read supported camera, date, location and AI fields in your browser.
No AI image detector is always right, including this one. Accuracy depends on the generator, the kind of image and what happened to the file after it was made. We don’t print a single headline percentage. Every result lists its evidence, and our methodology explains where each check fails. When benchmark numbers are published, they will include the false-positive rate on real photos.
A detector examines a file, not a person’s identity or intent. Even strong AI evidence does not establish who generated the image, who uploaded it, or whether they meant to mislead anyone.
No signs of AI means this check found no AI evidence strong enough to flag. An image may still have been edited, staged or used out of context. Missing metadata is common after ordinary sharing and does not make a photo fake.
The check does not perform reverse image search or fact-checking. A newly generated face may have no matches elsewhere, while a real photo can still be stolen, miscaptioned or shown out of context.
Yes. You can check one image online a day without an account. A full report is sold as one-time credits, from $2.99, with no subscription.
Choose the image above. We look for supported signs of AI generation, AI edits and other processing, and explain what the check found and what remains uncertain.
It can find supported editing traces and AI labels. A software name shows that an app saved the file; it does not establish what changed. AI detection can miss small edits, especially when labels have been removed.
Yes. JPG, PNG, WebP and HEIC files up to 20 MB are supported. Upload the original file instead of a screenshot for the best result.
Yes. Download the image and choose it above, or paste a supported direct-image or X post link. A screenshot also works, but it often loses file details. Use the original when you can.
No detector is always right. We show the evidence behind each result and never present a guess as proof. Read how the method works and where it fails.
We do not keep a server-side image library or train models on your images. AI detection sends an image copy to the providers listed in our Privacy Policy. From phone uses temporary transfer storage, and checkout can temporarily save the file on your device so your report can resume. The policy explains these cases and provider terms.
It can read supported content credentials when they remain in the file. Current supported OpenAI images can also contain SynthID, which we do not decode. OpenAI’s own verification tool can check supported OpenAI provenance signals. Missing signals do not rule out AI.
Yes. Google offers SynthID verification for supported images from Google and participating partners, and origin information through some Search and Gemini features. We do not decode SynthID. Our AI watermark detector page explains what we can read and links to the official tools.
Labels and credentials are lost when an image is screenshotted or re-saved, and heavy editing can fool pixel-based detectors. That is why each result shows how strong its evidence is. We do not offer tools to evade detection.
It checks still images for signs of AI generation and editing. A full report may include a face-swap signal where supported. It does not verify anyone’s identity, and it does not analyze video or audio.
Choose an image above. We read its content credentials, supported AI labels, generator settings and camera data. The free check uses up to one detector when needed; the full report attempts two. You see the findings, limits and suggested next steps.
Yes. If labels are missing, the free check can examine the pixels with one detector. The full report shows additional findings when available. Missing metadata alone is not evidence of AI, and screenshots or compressed copies can be harder to assess.
Yes. Digital art, illustrations and anime-style images can be checked the same way. Stylized images give detectors less to work with, so an Inconclusive result is common for them.
It can look for signs of AI in an image of a document, but detection can be wrong. It cannot authenticate an ID, receipt or document. Verify important details with the issuer.
Sources and official tools
Work through seven practical checks, from file details and edit history to inconsistent shadows, bent lines, repeated textures and earlier copies online. Learn why several signs together are more useful than one suspicious detail.
Understand the difference between reverse image search and AI detection. TinEye can find older or uncropped copies, while an AI check examines origin evidence; using both answers more questions than either alone.