Image Authenticity Checker

Detect image manipulation, copy-paste cloning, noise inconsistencies, and AI-generation metadata — entirely in your browser

Drag and drop files here or click to select

JPG, PNG, WebP, BMP — one image, up to 20 MB

Error Level Analysis

Clone / Copy Detection

Noise Consistency Map

Noise Naturalness Map

Red = low noise kurtosis (AI-like). Green = natural photo statistics.

AI Generation Detection

Other tools you may like

About Image Authenticity Checker

With AI-generated imagery and photo manipulation becoming increasingly accessible, verifying image authenticity is more important than ever. This browser-based tool applies four forensic techniques to help identify potential manipulation: Error Level Analysis (ELA) to detect re-edited areas, clone/copy detection to find copy-pasted regions, noise consistency analysis to spot composited regions from different sources, and metadata forensics to identify AI generation tags or missing camera information.

All analysis runs entirely in your browser using JavaScript and the HTML5 Canvas API. No image data is transmitted to any server. Results are probabilistic indicators — not definitive proof of manipulation.

How to Use?

Click the upload area or drag-and-drop an image (JPG, PNG, WebP, or BMP).
The tool automatically runs four forensic analyses. Wait a few seconds for results.
Review the overall verdict and explore each analysis tab — ELA, Clone Detection, Noise Map, and Metadata — for detailed findings.

Notes:

ELA is most reliable on JPEG images. Lossless formats (PNG/WebP) produce higher baseline ELA values due to JPEG recompression, which can cause false positives.
Clone detection uses block fingerprinting. Uniform-color areas (clear sky, walls) may generate false matches. Results should be corroborated with other analysis tabs.
These techniques provide probabilistic clues — not forensic proof. Legitimate operations like sharpening, noise reduction, or social-media recompression can also trigger anomalies.

Frequently Asked Questions

What is Error Level Analysis (ELA)?

ELA re-compresses the image at a known JPEG quality level and highlights areas that deviate from the expected compression level. Regions that have been edited, copied, or pasted often compress differently from the original parts of the image, appearing as bright areas in the ELA visualization.

How does clone detection work?

The image is divided into small blocks. Each block is fingerprinted using its spatial color distribution. Blocks that are far apart but share nearly identical fingerprints are flagged as potential clone pairs — a common sign of copy-paste retouching used to remove or duplicate objects.

What does noise analysis reveal?

Every camera and image source produces a characteristic noise pattern. When an image is composited from multiple sources, the noise characteristics differ across regions. The noise consistency map highlights areas with unusually high or low noise compared to the rest of the image, which can indicate splicing.

Can this tool detect AI-generated images?

The metadata forensics scan checks for AI generation tags embedded by tools like Stable Diffusion, DALL-E, Midjourney, and others, as well as C2PA/Content Credentials. Missing camera EXIF data in a high-resolution JPEG can also hint at AI origin. Pixel-level detection of modern AI images remains challenging without specialized models.

Is my image uploaded to a server?

No. The entire analysis runs locally in your web browser using JavaScript and the HTML5 Canvas API. No image data is sent to any server — your content stays completely private.