Frequently asked questions

An AI fingerprint is hidden data embedded in AI-generated images that identifies them as AI-created. This includes EXIF metadata tags, C2PA content credentials, PNG text chunks with generation parameters, and pixel-level watermarks like Google's SynthID.

C2PA (Coalition for Content Provenance and Authenticity) is an open standard for embedding verifiable provenance data in media files. AI tools like DALL-E, Midjourney, and Firefly use it to sign images with certificates proving they were AI-generated. Platforms read these certificates to show 'Made with AI' labels.

Yes. Instagram reads C2PA metadata and IPTC DigitalSourceType tags embedded in images. If these indicate AI generation, Instagram displays a 'Made with AI' label. The detection is metadata-based, not visual.

Yes. okie.fun is a free, browser-local tool that strips EXIF, XMP, IPTC, C2PA, and PNG text chunk metadata from AI images. It also offers LSB perturbation to disrupt pixel-level classifier detection. No signup required.

The 'Metadata Only' method changes zero pixels — output is byte-identical to the original. The 'Combined' method applies LSB perturbation (max ±1/255 per channel), which is invisible to the human eye. PSNR stays above 80dB.

LSB (Least Significant Bit) perturbation randomly flips the lowest bit of selected pixel color channels. The maximum change is 1 out of 255 brightness levels — completely invisible, but enough to disrupt statistical patterns that AI classifiers look for.

No. All processing runs locally in your browser using the HTML Canvas API. Your images never leave your device. No data is collected, stored, or transmitted. See our privacy policy for details.

As of 2026: OpenAI (DALL-E 3, GPT-4o), Adobe Firefly, Google Gemini, Midjourney (since late 2024), and Leonardo.Ai. Stable Diffusion does not natively embed C2PA, but some hosting platforms add it.

SynthID is Google's invisible watermarking technology that embeds patterns directly into pixel values. Unlike metadata, it survives re-saving and format conversion. It's used in Google Gemini images. LSB perturbation can disrupt some SynthID patterns.

Screenshots strip all metadata (EXIF, C2PA, etc.) but significantly reduce image quality. A dedicated tool like okie.fun preserves full quality while removing the same metadata. Screenshots also don't address pixel-level watermarks.

Removing metadata from your own images is legal in most jurisdictions. However, laws vary by region and use case. okie.fun is a tool — how you use it is your responsibility. We recommend using it ethically and in compliance with applicable laws.

okie.fun supports PNG, JPEG, and WebP input formats. Output is available as lossless PNG or high-quality JPEG (97% quality). The output resolution matches the input exactly.

Use okie.fun's AI Detector tool. It scans for EXIF metadata, C2PA blocks, PNG text chunks, and statistical patterns in pixel noise. It gives a percentage score indicating how likely the image is to be flagged as AI-generated.

Pinterest has begun flagging AI-generated content in some regions, primarily by reading C2PA and EXIF metadata. Removing these markers with okie.fun prevents the AI label from appearing.

Metadata Only strips all metadata (EXIF, XMP, C2PA, PNG chunks) with zero pixel change — the safest option for quality. Combined mode does everything Metadata Only does, plus applies LSB perturbation to disrupt pixel-level classifiers. Combined is recommended for maximum protection.