AI Image/Video Origin Detection

AI Image & Video Detector: see what the file left behind

Drop in an original image or video. AICheck365 inspects the C2PA, EXIF, XMP, PNG text, generation parameters, and video-container origin clues.

Files are processed in your browser and never leave your device.

Drop an original file

Privacy first Files are only read inside your browser
Best with originals Original files retain more detectable signals
Explainable results Matched fields and confidence are listed

Important: This is not a magic AI detector. It only reads EXIF / XMP / C2PA / PNG text chunks / MP4/MOV metadata / file-name signals that still exist in the file. Video-frame analysis runs locally and may be slower.

sample.png├── PNG Header├── C2PA Manifest├── XMP CreatorTool├── digitalSourceType├── claim_generator└── Embedded Parameters

AICheck365 inspects the inside of the file, instead of guessing from the picture.

Upload Image/Video

Select an image or video and analyze it locally in your browser.

Click or drag an image/video here JPEG / PNG / WebP / GIF / BMP / TIFF / MP4 / MOV / M4V / WebM / AVI · multiple files supported

Initializing…

The detector needs JavaScript to analyze files locally. Even without JavaScript, the methodology, supported formats, and platform guides below remain readable.

Detection Results

Results list matched signals, confidence, field details, and limitations.

Expert mode

Try an original file. Less processed files keep more detectable signals.

Detectable Platforms / Tools

This only represents currently detectable trace patterns, not every AI image or video platform.

Supported Signal Types

C2PA Content Credentials / JUMBF
EXIF metadata
XMP / IPTC metadata
PNG tEXt / iTXt chunks
MP4 container metadata / AIGC labels / SEI markers
Sampled video frame watermark analysis
Filename heuristics

Supported Formats

Recent Detections

Click to review results stored only in this browser.

Why evidence disappears

Each share, compression, or screenshot may erase AI-origin signals.

Original platform file Intact

All metadata intact: C2PA, XMP, EXIF, PNG text chunks, generation params
Chat app transfer (WeChat / Telegram) Lost

Lost: PNG text chunks, most XMP fields, EXIF GPS data
Social platform transcode (Weibo / Twitter / Instagram) Lost

Lost: All EXIF fields, C2PA Manifest, XMP CreatorTool, all generation params
Screenshot or screen recording Lost

Lost: All original metadata; only screenshot tool EXIF remains
Re-export / save-as Lost

Lost: All provenance fields; new software may write unrelated EXIF

No signals found does not mean it is not AI-generated. Evidence may have been erased during sharing.

Five evidence layers

Click each layer to see example fields that are inspected.

File type / Basics

Format:          PNG / JPEG / WebP / MP4 / MOV
FileSize:        2.4 MB
ColorSpace:      sRGB
BitDepth:        8
ImageWidth:      1024
ImageHeight:     1024

EXIF + XMP metadata

Software:                DALL-E 3
Make:                    OpenAI
XMP:CreatorTool:         Adobe Firefly
XMP:DigitalSourceType:   trainedAlgorithmicMedia
XMP:CreateDate:          2024:03:15 14:22:07
XMP:MetadataDate:        2024:03:15 14:22:07

PNG text chunks + generation params

tEXt:parameters:   masterpiece, best quality, 1girl
tEXt:prompt:       a futuristic city at night
tEXt:negative:     lowres, bad anatomy, blurry
tEXt:steps:        20
tEXt:sampler:      DPM++ 2M Karras
tEXt:model_hash:   d8691b4d18

C2PA + digital signature

c2pa:claim_generator:   Adobe Firefly 1.0
c2pa:claim_generator_hints: adobeFirefly
digitalSourceType:      trainedAlgorithmicMedia
c2pa:title:             Firefly_20240315.jpg
Signature:              VALID (SHA-256, RS256)
Certificate:            Adobe Inc. / DigiCert

Video container + SEI + frames

Handler:         kling-ai-1.6-standard
Encoder:         Kling Video v1.6
UserData:        {"generator":"kling","model":"v1.6"}
SEI:UUID:        kling_provenance_v1
FrameWatermark:  detected (confidence: 0.91)
MetadataTrack:   sora_video_metadata_v1.0

Researched Platforms

Dedicated detection guides exist for the platforms below. Click to see common origin signals.

What does this tool actually check?

AICheck365 looks for source signals preserved in image and video files: EXIF, XMP, C2PA, PNG text chunks, MP4/MOV metadata, file names, and sampled video-frame watermarks.

It can recognize common traces from Gemini, Google Imagen, DALL-E / GPT Image, Midjourney, Stable Diffusion, Adobe Firefly, Kling, Sora, Runway, Pika, and similar platforms.

Platform guides · AI video detection guide

How to interpret results

Detection is based on metadata and container signals, not pixel guessing
Social-media compression or transcoding often strips metadata
Screenshots, screen recordings, and re-exports lose original signals
Results are useful for quick triage, not legal attribution

Methodology & Limitations

We can check

File metadata (EXIF / XMP / IPTC)
Provenance credentials (C2PA / Content Credentials)
Digital signatures and hashes
Platform marks and generation parameters
Container structure and file headers
Some edit and export clues
PNG text chunks
MP4/MOV container metadata and SEI markers

We cannot guarantee

All AI files retain provenance data
Screenshots retain origin information
All platforms use uniform marks
No signal means human-made
Results replace legal or expert appraisal

Evidence beats intuition. Transparency beats percentages.

Read the full methodology →

Real Sample Experience

These samples come from real tests. Click to analyze in the detector.

Adobe Firefly image with valid Content Credentials Adobe Firefly

JPEG Unprocessed 2026-06-19

Source: Content Authenticity Initiative example assets (contentauth/example-assets, MIT)

1 signals: C2PA

Known limits:
digitalSourceType = trainedAlgorithmicMedia and the signature is cryptographically valid.
The signing certificate is not in this engine's trust list (untrusted), so the result is reported as 'valid' rather than 'trusted'.

AI-generated video with valid Google C2PA Content Credentials Google Veo

MP4 Unprocessed 2026-06-19

Source: User-provided original video carrying intact C2PA (Google generative-AI provenance)

2 signals: C2PA (Google Generative AI), MP4 encoding tool: Google

Known limits:
digitalSourceType = trainedAlgorithmicMedia (fully AI-generated); the BMFF C2PA manifest records 'Created by Google Generative AI' plus an imperceptible SynthID watermark, and the signature is cryptographically valid.
The manifest identifies 'Google Generative AI' (Veo is Google's video model); the signing certificate (Google LLC) is not in this engine's trust list, so the result is reported as 'valid' rather than 'trusted'.

OpenAI image whose C2PA certificate has expired OpenAI (ChatGPT / DALL·E)

PNG Unprocessed 2026-06-19

Source: Content Authenticity Initiative example assets (contentauth/example-assets, MIT)

1 signals: C2PA

Known limits:
A C2PA claim marked trainedAlgorithmicMedia is present, but the signing certificate is expired and untrusted.
Validation therefore fails: the provenance claim exists but cannot be verified (evidence state D).

Stable Diffusion XL output with embedded generation parameters ComfyUI / Stable Diffusion XL

PNG Unprocessed 2026-06-19

Source: ComfyUI official examples (comfyanonymous/ComfyUI_examples)

1 signals: PNG text chunks (prompt / workflow)

Known limits:
No cryptographic signature: only unsigned generation parameters embedded in PNG tEXt chunks.
Such chunks are routinely stripped by chat apps, social platforms, and screenshots.

Image with no provenance metadata —

JPEG Unprocessed 2026-06-19

Source: C2PA Rust SDK conformance fixtures (contentauth/c2pa-rs, Apache-2.0 / MIT)

Known limits:
No origin signals were found. This is not proof of human creation.
Evidence may have been removed earlier in the file's distribution chain.

Plain photograph, no embedded provenance —

JPEG Unprocessed 2026-06-19

Source: NASA (public domain), via contentauth/c2pa-rs fixtures

Known limits:
No AI or provenance signals detected.
Absence of signals cannot distinguish a real photo from an AI image whose evidence was stripped.

Tampered C2PA signature (content hash mismatch) C2PA test sample (tampered)

JPEG Processed 2026-06-19

Source: C2PA Rust SDK conformance fixtures (contentauth/c2pa-rs, Apache-2.0 / MIT)

Known limits:
A C2PA manifest is present but the asset hash does not match (assertion.dataHash.mismatch) and the certificate is untrusted.
The provenance claim is therefore invalid (evidence state D).

Latest Experiments

Real batch test results.

Adobe Firefly 2026-06-19

JPEG 1 samples

Valid C2PA Content Credentials with digitalSourceType=trainedAlgorithmicMedia; the signature is cryptographically valid while the signing certificate is untrusted in the browser trust list.