Library

ImageVerity blog and guides

Practical articles about AI-image review, provenance, metadata, detector signals, and real operational workflows.

Start here

Core guides

Foundational articles that explain the main verification layers and how to use them together.

Foundation Guide

How to tell if an image was generated by AI

A practical workflow for reviewing provenance, metadata, visual inconsistencies, and detector signals together.

Read article

Foundation Metadata

Can EXIF metadata detect AI images?

Where metadata helps, where it fails, and how to use it correctly in a broader review workflow.

Read article

Foundation C2PA

What C2PA can and cannot prove

A practical guide to using content credentials without overclaiming what provenance means.

Read article

News analysis and strategic context

Editorial pieces that connect current provenance updates to real operational review decisions.

Editorial illustration for verifying suspicious AI images after provenance updates

News analysis Jun 12 · journalists

How newsrooms should verify suspicious AI images after provenance updates

How newsrooms should verify suspicious AI images after provenance updates, and what provenance, EXIF, detector scores, and manual review each can and cannot prove.

Read article

Editorial illustration about verifying AI images with SynthID and C2PA across Search and Chrome

News analysis Jun 13 · journalists

What Google's SynthID and C2PA expansion means for verifying AI images

Google's May 19, 2026 transparency update expands SynthID and C2PA verification into Search and Chrome. Here is what that changes for AI-image verification in real workflows.

Read article

Deep dive

Workflow explainers

More specific explainers for teams refining review standards, escalation rules, and evidence handling.

Illustration for understanding why AI image detectors can be wrong

Workflow Jun 10 · journalists

Can AI image detectors be wrong?

AI image detectors can be wrong. This guide explains false positives, false negatives, and the review workflow teams should use instead of trusting one score.

Read article