Where AI helps with content, where it harms, and how to tell the difference on a small WordPress site.
AI for content is a real tool with a real risk. Used as a research assistant, a first-draft engine, or a copy editor, it speeds the work up. Used as the publisher, generating dozens of posts that nobody at the business has lived through, it is the fastest way to a Google indexability problem we have seen this decade. Our articles cover both sides honestly. We use AI ourselves where it helps. We refuse it where it hurts.
The pattern that recurs across every AI-content recovery we have run is the same: the volume looks like progress, the impressions counter ticks up, and then the next core update lands and most of the new pages stop ranking. The fix is not more AI. The fix is the slow honest publishing rhythm and the structural layer underneath it. If you are weighing up an AI content plan for your site, or trying to undo one, these articles are written for you.
Saturday 30 May produced the second weekend of severe ranking volatility in Google's May 2026 core update. SEOs are reporting 50 to 100 percent traffic drops, with AI-built commodity pages absorbing the bulk of the damage. Here is what client GSC reports are actually showing this morning, and the recovery playbook drawn from a real UK case.
A small UK removals website published a year of AI-generated "moving tips" posts. Google now refuses to index 220 of its 272 pages. Here are the GSC screenshots, what E-E-A-T actually means, what non-commodity content looks like in practice, and a five-phase recovery playbook.
Press Forge
