Google confirmed on September 25, 2026 that the September 2026 spam update is live, and the rollout will take roughly two weeks — not the two days a typical adjustment takes. This is a confirmed statement from Google, not tool-signal speculation. The update applies globally and affects all languages, but Google says it only targets sites that spam.

Forum trackers expect ranking volatility over the full rollout window. That expectation comes from community observation, not from Google. Webmasters who see drops before mid-October should resist early conclusions.

What does the spam update change?

Google distinguishes this release from a core update. A spam update targets specific policy violations rather than recalibrating overall quality assessment. It is the fourth spam update Google has shipped this year, according to Search Engine Roundtable's count.

Google has not published per-vertical impact data. It rarely does for spam updates. Sites in heavily spammed verticals — pharma, finance, ticketing — historically show the largest swings, but that pattern is inference from prior updates, not from this rollout. The confirmed facts are scope and duration: global, all languages, about two weeks.

Takeaway: this update punishes spam, and Google says nothing about reward for clean sites.

How did OpenAI's quality raters misbehave?

A report by 404 Media found that OpenAI's AI trainers — the equivalent of Google's thousands of quality raters — cheated by using AI to complete their work. Google uses human raters to test and improve search results and AI outputs; OpenAI runs a parallel program, and the report says its trainers subcontracted the job to the models themselves.

The detail matters for anyone weighing AI output quality claims. If training and evaluation data carries AI contamination, downstream benchmarks inherit the flaw. OpenAI has not, per the report, fully addressed how it screens for this. The absence of a stated fix is itself information.

Takeaway: the people grading the AI reportedly let the AI do the grading.

What changed in Google's local and ads products?

Three confirmed product changes surfaced the same day.

• Local Services Ads tracking. LSAs now show the Google Business Profile phone number even when advertisers instruct Google to use the Google Ads number. Sources describe this as a confirmed change. Google has not reverted it, and nobody has said whether it will.
• Google Posts policy. Google updated two help documents for Google Posts. Phone numbers are now allowed only when they are verified contact information. Unverified numbers are prohibited.
• Google Ads limitation. Advertisers cannot edit Target ROAS while promotion mode is active. Google displays the notice: "You can't edit your Target ROAS while promotion mode is active," and offers ROAS tolerance editing instead.

The LSA change has direct budget consequences. Call attribution that routes through GBP numbers instead of Ads numbers can break offline conversion tracking. Google has not published a timeline for a fix. Advertisers running call-based campaigns should audit their attribution data now.

What did Bing and Google schema announce?

Microsoft is testing two interface changes. Bing is testing editable, customizable icon labels on images within search, powered by Bing's AI image feature. Separately, Microsoft is testing a rename of the Copilot search tab in Bing to "AI Mode" — the same label Google uses for its AI search experience. Both are tests. Tests in Bing search do not always reach general release.

Google updated its video structured data documentation. The VideoObject schema now documents support for the creator and author properties, and Google clarified the supported interaction types for interactionStatistic. Google's own wording: the updates serve "to document support for the creator and author properties, and clarify supported interaction types for interactionStatistic in VideoObject structured data." Publishers using video markup should check whether their current markup exposes creator data, since documented support often precedes rich result display changes — though Google has promised no such display change here.

What should observers watch next?

Two independent commentators flagged data-quality warnings worth noting. Barry Adams wrote on LinkedIn that reports citing Chartbeat data claiming publishers' Google traffic fell 40 percent year on year deserve "a huge bag of salt." Joy Hawkins explained on X why GBP "searches breakdown" queries often appear unrelated to a business's actual brand or services — proximity visibility is one common cause.

Watch the spam update through its expected completion in early October, watch whether Google restores LSA call tracking to the Ads number, and watch whether Bing's "AI Mode" tab rename graduates from test to launch.