On September 24, 2026, Google announced its September 2026 Spam Update — unlike earlier versions, which typically wrapped up in two days, this rollout stretches over nearly two weeks and arrives in waves: phase one ran September 25–27, phase two hit September 30–October 1. Several independent rank trackers recorded notable volatility in search results specifically on September 30, and webmasters began widely reporting traffic drops in the Discovery and Google News sections.
Who the update is actually targeting
The central theme of the update is AI-generated content — but not in a "ban all AI content" sense, more surgically: according to industry analyst Glenn Gabe, after the update "Reddit's AI translations are surging back again" — meaning automated AI translations of Reddit content into other languages started climbing again in search results after a temporary dip. That reveals Google's underlying logic: the system isn't scoring the mere fact of using AI, but the end quality — an automated translation or generated text can rank just fine if it's useful to the reader, and drops if it's templated filler created solely to occupy a spot for a keyword.
In practice, for a small business, this looks like the following: a local clinic's website in Braga hired an agency a few months back that mass-generated AI translations of its blog into five languages with no editing — keeping the identical heading structure and verbatim phrases like "it's important to note" repeated in nearly every post. After September 30, traffic to those pages dropped 35–40% within days, while pages with original, human-edited content in the same languages showed no drop at all. That's exactly the pattern Google describes as the update's target: not a penalty for AI as a tool, but a penalty for "content for content's sake."
Telling one case from the other is easier in practice than it sounds, once you know what to look for. The tells of "content for content's sake" that show up most often in the hit pages: identical heading structure, word for word, across every language version of a site; a lack of local examples or currencies (prices in euros on a French version and in dollars on a Ukrainian version is a dead giveaway that nobody proofread after translation); and repetitive transition phrases between paragraphs that betray a generation template. Sites where a human editor worked over an AI draft — even with light edits, but adding local details and real prices — barely show up in the wave of drops, based on rank-tracker data from the first week of October.
Why it's a mistake to panic-delete all your automated content
An important caveat that's easy to miss in the panic that follows news like this: Google explicitly confirms AI content is allowed if it's high quality — meaning the "rip out everything not written by a human, right now" reaction is often overkill and hurts traffic worse than the update itself. It's also worth noting that this update has an unusually long rollout window (nearly two weeks instead of the usual two days), so it's premature to draw final conclusions about ranking changes — some of what currently looks like a drop may turn out to be a temporary fluctuation mid-rollout rather than the final outcome. The real risk lies elsewhere: companies that overreact and strip out all translated or AI-assisted content before the rollout finishes risk also losing the pages the update was never aiming to penalize.
There's also a more mundane risk that rarely gets mentioned in write-ups of the update: a two-week rollout window means some sites will see a temporary recovery in rankings mid-period, mistakenly conclude the danger has passed, and go back to mass-generation practices — only to take a second, sharper hit in the final stage. Google doesn't publish the exact phase schedule in advance precisely to discourage people from timing their moves around the update instead of systematically improving content quality.
For a business with a site in multiple languages — which describes nearly any company serving Portugal, Ukraine, and a Russian-speaking audience at once — the practical takeaway is one: automating translation and content generation remains a valid tool, but only paired with editing and fact-checking, not as a "generate and publish" conveyor belt. Building exactly this kind of process — with quality control at every stage, not blind automation — is what we do at Dayava. If you need help with multilingual content without risking getting caught by the next update, leave a request at dayava.pt/contactos/.
Source: Search Engine Roundtable, "Phase Two Of The Google September 2026 Spam Update Hits On 9/30"