Google September 2026 Spam Update: Scaled Content in Focus

What happened

Google has started rolling out its September 2026 spam update, and the company says the full deployment should take roughly two weeks to finish 1. As with previous spam updates, the announcement was short on detail. Site owners will spend the next couple of weeks watching their traffic dashboards and trying to work out what changed.

The early picture is clearer than usual on one point: scope. According to reporting on the release, Google confirmed that this update does not target link spam. Its focus is scaled content abuse 2. That is Google's term for producing large volumes of low-value pages, often with automation, mainly to rank rather than to help readers.

The AI-slop narrative, and where it overreaches

The update arrived during a wider industry worry about AI-generated "slop" flooding search results. Much of the commentary has linked the rollout to a Google system called SAFE (Scaled Abuse Forensics Examiner). The evidence for that link is weak. SAFE comes from a May 2026 research paper about clusters of YouTube channels, and there is no evidence Google uses it for Search ranking at all 2.

This matters because SEO communities tend to turn research papers into assumed ranking factors. A paper describing how Google can detect coordinated abuse on YouTube shows Google has the capability and interest. It does not show the same model now judges web pages. Treating SAFE as the engine behind this spam update is inference, not confirmed fact 2.

What Google's own people are saying

The more reliable signals come from Google's public statements. At the Search Central Live Deep Dive event in Barcelona, Gary Illyes reportedly said scaled AI content has become a bigger problem than link spam 2. That is a notable change in emphasis. For much of search's history, manipulative linking was the main spam fight, and Google built whole update cycles around it.

A recap of the Barcelona event by John Campbell adds that Google is explicitly using AI to catch AI slop at scale 2. The logic is simple. When generative tools make it nearly free to publish thousands of pages, manual review and older heuristics cannot keep up. Detection has to be automated and has to work at the same scale as the abuse.

These comments fit with the stated scope of the September update. Google says the release targets scaled content abuse rather than links 2. Its representatives say scaled AI content is the larger threat. Read together, the update looks like a deliberate move toward the problem Google now considers most pressing.

Beyond text: the video front

Google's anti-slop work is not limited to written pages. The company has also published a separate synthetic-media detection system aimed at coordinated AI video spam 2. It was tested across 16,250 channels on a weekly basis and reported a false-positive rate below 0.05% 2.

That false-positive figure is the most important number in the story. The biggest risk in cracking down on AI-generated content is catching legitimate creators who use automation responsibly or who simply produce at high volume. A sub-0.05% error rate in testing suggests Google is aware of that risk. Lab performance does not always hold up in production, though, and the system concerns video, not Search.

What it means for publishers

Here is how the pieces fit together, as analysis rather than confirmed mechanism:

• The confirmed facts are narrow. A spam update is rolling out over about two weeks 1, and Google says it targets scaled content abuse, not links 2.
• The broader context is real. Google staff are publicly framing scaled AI content as a top concern, and the company is investing in AI-driven detection across formats 2.
• Specific attributions are speculative. Linking ranking drops to SAFE or any other named research system goes beyond the available evidence 2.

Sites most exposed are likely those that have published large amounts of templated or lightly edited machine-generated pages built mainly for search traffic. Sites that use AI as a drafting aid but add original reporting, expertise, or real editorial value fit Google's stated concern less well. Even so, nobody outside Google can say exactly where the line falls.

Publishers should hold off on reacting until the rollout finishes. Rankings often move around mid-deployment, and changes made before the dust settles can be hard to evaluate. Once the update is complete, sites hit hard should audit for thin, mass-produced content rather than chase theories about a particular detection model.

The broader point is that Google's spam priorities have shifted. Link schemes are no longer the main concern. Content produced at industrial scale is now the main target, and Google is building automated tools to deal with it.