AI Image Detector Core Features

Four Independent Detection Models

The AI image detector runs Sightengine GenAI, Winston AI, Was It AI, and an open-source SAFE engine in parallel, each casting a weighted vote under a published rule.

C2PA Credentials Trump Card

C2PA credentials and generator self-tags are checked first. When certain, they override every model vote, providing deterministic file-level evidence.

Published Weighted-Vote Rule

Every report attributes results by name. The weighted-vote combination produces one of five grades, with engine disagreement reported honestly.

Batch Screening

Pro and Team users can screen up to 50 images at once, receiving a triage summary and CSV export for efficient moderation workflows.

Shareable Report Links

Each analysis generates a shareable report link with forensic PDF download, useful for moderators, educators, and editorial teams requiring documented evidence.

SHA-256 Chain of Custody

The AI image detector records the SHA-256 hash of each uploaded image, supporting chain-of-custody documentation for research and editorial purposes.

Credit Refund on Engine Failure

When a detection engine fails, its credit automatically refunds. The report still runs with the remaining engines, ensuring users only pay for successful analyses.

• Dating and marketplace safety: Verify profile photos before trusting them, with shareable report links for moderators.
• Educators: Carry integrity conversations with graded AI detector reports rather than bare percentages.
• Artist communities: Screen submissions to keep human-made galleries human, with batch triage for mod queues.
• Editorial and research: Document image provenance before publishing, with SHA-256 hashes for chain-of-custody notes.
• News verification: Check viral images through the AI image detector before sharing on social media.