Content authenticity signals

Content authenticity is the set of signals that tell a reader, a search engine or an AI answer engine where a piece of content came from and how it was made. Some of those signals live in the visible page: a byline, a review with a real customer's name, a caption that says an image was generated. Others live in markup and metadata that machines read directly. This section of the handbook covers the second kind, and how to keep it honest and consistent with the first.

Why authenticity is a markup problem

Structured data was built to describe pages: this is a recipe, this is a product, this is a review with four stars. Once machines read those labels, the labels become claims. A review rating in JSON-LD is a claim that a real person rated the item. An author property is a claim about who wrote the article. An embedded metadata field in an image can be a claim that the picture came from a camera or from a generative model.

Generative tools made those claims cheaper to fake and harder to check by eye. Regulators, search engines and standards bodies responded in different ways, and each response touches markup or metadata somewhere. The guides below follow those touch points one at a time, with the primary source linked for every rule, and as of November 2026 none of them promises a ranking, a star or a citation for adding a label.

Reviews you can trust

Reviews were the first place where structured data and authenticity collided. Review snippets only make sense when the reviews behind them are genuine, and both search engine policy and consumer protection law now say so in writing. Our guide to AI-written reviews and review snippets sets out what counts as a fake review, what Google's review snippet guidelines require, what the 2024 FTC rule on fake reviews prohibits, and how to screen reviews you collect before any of them reach your markup.

Disclosing AI help

Most publishers now use AI somewhere in their workflow, from research notes to image edits. The question is what to tell readers and what, if anything, to say in markup. Disclosing AI-assisted content covers what Google has published about AI content and disclosures, the one schema.org property built for describing a digital source, how bylines and reviewers should work when AI helped, and a disclosure pattern that keeps the visible page and the JSON-LD saying the same thing.

Content Credentials

Content Credentials are the C2PA standard for attaching signed provenance records to images, video, audio and documents. They are not schema.org markup at all: they travel inside the file or alongside it. That makes WordPress image handling the main risk, because resized copies are new files. C2PA Content Credentials on WordPress explains what WordPress documents about metadata on resize, how to keep or link a manifest, how to tell readers that credentials exist, and how to check them after upload.

Image metadata labels

The simplest machine-readable label for an AI image is a single embedded field from the IPTC photo metadata standard. IPTC digital source type for AI images lists the field's values for generated and AI-edited images, explains the difference between the two values people confuse most often, shows how Google reads the field, and walks through checking a file before and after WordPress processes it.

How to choose a guide

Start from the thing you publish. If you collect or display customer reviews, begin with the reviews guide, then read the review snippet basics in the Schema types section. If you publish articles written or edited with AI help, start with the disclosure guide. If you publish images that were generated, edited with generative tools or signed by a camera or editing app, read the IPTC guide first, because that field is the label most tools can write today, then the Content Credentials guide for signed provenance.

Each guide names its sources, dates what it checked and says plainly where the documentation is silent. Where a standard and a search engine disagree, or where a search engine has not said how it uses a signal, the guide says that too rather than guessing. This section is part of wprichsnippets, an independent handbook on structured data for WordPress and AI answers, and it will grow as the standards and the search engines' documentation change.