Dev.to WebDev πŸ›  Dev πŸ‘ 0 πŸ“– 2 min read

How I label AI-written posts in WordPress (and prove a human reviewed them)

The EU AI Act's Article 50 transparency obligations have applied since 2 August 2026. If you run a WordPress site that publishes AI-assisted content, readers (and soon regulators) will expect a clear answer to one questi

The EU AI Act's Article 50 transparency obligations have applied since 2 August 2026. If you run a WordPress site that publishes AI-assisted content, readers (and soon regulators) will expect a clear answer to one question: what did AI write here, and did a human check it?

I built Labelwise, a WordPress plugin that answers that question per post. Here is how it works and the design decisions behind it.

Not legal advice. Labelwise helps you show and document what you disclose β€” it does not guarantee compliance.

1. Four honest labels, chosen by a human

Labelwise does not try to detect AI text. Detectors are unreliable, and the decision belongs to your editors. Each post gets one of four states:

State Badge text (editable)
none no badge
ai_assisted AI-assisted content
ai_generated AI-generated content
human_reviewed AI-assisted, reviewed by a human

The badge is a plain, accessible <aside role="note">, inserted above or below the content, or placed anywhere with the [labelwise_badge] shortcode. Light and dark styles, no tracking, no external requests.

2. A publishing gate for unreviewed AI content

Article 50(4) has an interesting exception: AI-generated text on matters of public interest must be disclosed unless it has gone through human review and someone holds editorial responsibility. So the review step matters.

Labelwise Pro hooks into the publish transition: if a post labelled ai_assisted or ai_generated is published, it is held as Pending until an editor marks it as human-reviewed. Simplified:

if ( in_array( $status, array( 'ai_assisted', 'ai_generated' ), true ) ) {
    wp_update_post( array( 'ID' => $post->ID, 'post_status' => 'pending' ) );
}

Every label change goes into a review log: who, from what, to what, and when.

3. Machine-readable markers

There is no settled standard yet for marking AI text on the web, so the plugin exposes the label in three simple, documented places that tools can read today:

<meta name="labelwise:ai-disclosure" content="human-reviewed" />
"labelwise": { "status": "human_reviewed", "ai_involved": true, "reviewed": true }
<labelwise:status>human-reviewed</labelwise:status>

(HTML <head>, the REST API field on each post, and an RSS element under the urn:labelwise:ai-disclosure:1 namespace.)

4. Chatbots too

Article 50 also covers people talking to AI systems. A [labelwise_chatbot_notice] shortcode puts a short notice next to your chat widget.

Try it

I'd love feedback, especially on the markers: if a standard for AI-content metadata emerges, what should the plugin output?

This post was drafted with AI assistance and reviewed by the author.

πŸ“° Read the original article on Dev.to WebDev

Originally published by Dev.to WebDev. Aggregated on AIWithGhost for educational purposes β€” full credit and traffic to the original publisher.