Is it real? Better question:
what can we verify?

Most viral misinformation isn't a deepfake — it's a real photo with a false caption. Verity checks what evidence actually exists about a piece of media, and shows you its work. Three transparent verdicts, never the word “fake”.

Check media on Telegram Get the browser extension GitHub (Open Source)
1 media checked 0 verified 0 suspicious

The verdicts

Honest answers, not hot takes

Verified

A valid cryptographic signature (C2PA Content Credentials) proves who produced this file. Nothing else earns a check.

Unverified

No provenance found. That's the normal baseline of most web media — it means we don't know, not that it's false.

Suspicious

Evidence contradicts the media: a tampered signature, prior sightings under contradictory claims, or debunked fact-checks.

The evidence

Every verdict shows its work

Cryptographic ProvenanceC2PA Content Credentials — the only standard that can definitively verify origin.
Prior Sightings & HistorySHA-256 hashes and archive snapshots detect recycled footage repurposed with false narratives.
Metadata & ForensicsCamera EXIF, editing history traces, AI generator signatures, and GPS vs claimed geolocation.
Fact-Checks & News CoverageOptical character recognition (OCR) cross-checked against ClaimReview databases and GDELT.
Perceptual Hash Index (BK-Tree)64-bit pHash matching catches crop, re-encodes, resizes, and compression variations.
Optional Neural ClassifierSandboxed ONNX model can flag synthetic AI artifacts — it can never verify provenance.

Direct Answers • AEO & GEO

Frequently Asked Questions

What is Verity and how does it differ from AI detectors?

Traditional AI detectors guess based on pixel statistics and suffer from extreme false-positive rates on real camera photos. Verity is a multi-signal provenance engine. It evaluates tamper-evident cryptographic credentials (C2PA), reverse-lookups historical sightings in a BK-tree index, inspects forensic metadata, and searches journalist fact-checks.

Why does Verity never label an image or video as "fake"?

The vast majority of viral online misinformation is not generative AI deepfakes — it is authentic, unaltered media presented with fabricated context, deceptive dates, or misleading captions (known as cheapfakes). Calling media "fake" is scientifically inaccurate and breeds cynicism. Verity reports transparent evidence: Verified, Unverified, or Suspicious.

What is C2PA and how does Verity verify Content Credentials?

C2PA (Coalition for Content Provenance and Authenticity) is the global open standard supported by Adobe, Microsoft, Google, BBC, and Nikon. It cryptographically binds author, camera, and editing details to the file with an x509 certificate chain. Verity decodes and verifies these claims entirely in your browser using WASM.

How does Verity protect user privacy?

Zero media bytes are ever uploaded or retained. The browser extension and Telegram bot run media decoding and forensic parsing locally. Only mathematical SHA-256 digests and perceptual hashes are queried against the registry.

Integration & Deployment

Check media where it spreads

Telegram Bot

Forward any photo or video to @CheckVerityBot — receive an evidence-backed verdict and a shareable verification link in seconds. Free, zero installation.

Chrome & Firefox Extension (Manifest V3)

Right-click any web image → "Verify with Verity". Build from source: npm install && npm run build, then load packages/extension/dist at chrome://extensions. Chrome Web Store listing release in progress.

Prove Your Own Media (Signer CLI)

Content creators, photographers, and news organizations can stamp tamper-evident C2PA credentials directly into originals: npm run sign -- photo.jpg signed.jpg --cert cert.pem --key key.pem

Self-Hostable Registry API

GET https://verity.codemintah.dev/api/verdicts/<sha256> · GET /api/similar?phash=… · GET /api/stats — zero dependencies, lightning fast, privacy-preserving.