AI detection & provenance tools directory
Last updated October 10, 2026
No single tool answers "is this real?". This directory lists the main types of AI detection and provenance tools, what each one proves, and what it misses — so you can combine them correctly.
| Tool type | Examples | What it proves | What it misses |
|---|---|---|---|
| Watermark checkers | Google SynthID Detector | A file came from a participating generator that embedded a watermark. | Content from models that don't watermark; a clean result isn't proof of authenticity. |
| Provenance viewers (C2PA) | Adobe Content Credentials, VerifyAI C2PA verifier | Signed history of which device/app made and edited a file. | Files without a manifest — most social media uploads. |
| Forensic multi-modal detectors | VerifyAI | Signals of generation or editing across images, video, audio, PDFs and text, with heatmap and reasons. | A probability, not certainty; needs human judgement on amber results. |
| Text-only AI detectors | GPTZero and similar | Statistical likelihood that text was machine-written. | Images, audio and video; false positives on formal writing. |
| Reverse image search | Google Lens, TinEye | Whether an image appeared online earlier or elsewhere. | New generated images with no prior copies. |
| Evidence & custody tools | VerifyAI certificates | That a specific file (by SHA-256) existed unchanged at a given time. | Whether the content itself is truthful. |
Which to use when
- Payment, claim or hiring decision: forensic check + call-back on a known channel + keep a certificate.
- Publishing a news image: reverse search + C2PA + forensic check.
- Proving your own work: export with Content Credentials and issue a certificate — see for creatives.
- Training a team: the free workshop kit and quiz.
Related: VerifyAI vs IsThisAI · VerifyAI vs GPTZero · SynthID vs forensics.