What Did Google Change in Its Content Guidance on 1 October?

What the changelog says

Google keeps a public log of significant changes to its Search Central documentation. On 5 October its entry for 1 October read, in full: “Updated guidance on using generative AI content. What: Updated the using generative AI content guide with information from the Search Quality Raters guidelines. Why: To get our documentation in sync with our presentations we use at our developer events.” The developer events are the Search Central Live series; the latest, a three-day event in Barcelona, ended on 2 October, and Google speakers discussed quality, scaled content and AI-generated spam.

Search Engine Journal reported both changes on 1 and 2 October. Because the second change was not in the changelog, we compared the two pages ourselves.

The generative AI page: a fact-check instruction

Google's “guidance on using generative AI content on your website” is short. In June it ran to 343 words. On 5 October it ran to 397. The whole change sits in its section headed “Focus on accuracy, quality, and relevance”.

Three sentences were inserted after the section's opening line: “Keep in mind that generative models don't retrieve facts, but predict a likely sequence of words based on their training data. Because of this, generative AI outputs may contain inaccuracies (also known as hallucinations). It is critical to manually factcheck and review all AI-generated content for accuracy and trustworthiness before publishing.”

The next sentence changed by one phrase. In June it said the advice on accuracy “includes metadata like” title elements, meta descriptions, structured data and image alt text. It now says “This review also applies to metadata like” the same four. In other words, the hand check covers the parts of a page that can appear in search results, not only the article text.

Nothing else on the page changed. It still says generative AI can be useful for research and structure, still warns that generating many pages without adding value may breach the spam policy on scaled content abuse, and still says rater ratings “don't directly influence ranking”.

The people-first content page: an unlisted rewrite

Google's page on “Creating helpful, reliable, people-first content” is the one most often cited when people discuss how Google judges quality. It shows a last-updated date of 1 October, but has no changelog entry. Against our June copy, nothing was removed. Two passages were added.

Main content. A new section after the explanation of E-E-A-T says the rater guidelines treat the quality of the main content as “one of the most critical factors” in assessing a page. It defines main content as “any part of the webpage that directly helps the page achieve its purpose”, which can include text and media, tools, user contributions where they serve the page's purpose, tabbed sections, and titles and headings. It then lists the four attributes raters are trained to evaluate. Effort is “the extent to which human work went into creating the content”; Google gives original analysis as high effort, and pages generated from feeds or large amounts of AI text “without manual oversight or curation” as little to none, adding that “attribution or giving credit to other sources doesn't replace the need for original effort”. Originality means information or perspectives “that aren't already available on other websites”. Talent or skill is the expertise needed for a satisfying page, though “not all content requires specialized expertise”. Accuracy means informational pages should be factually accurate, and pages on topics that affect people's lives or wellbeing “must be highly accurate and consistent with established expert consensus”.

Deceptive authorship. The page has long encouraged bylines and author information. Directly after that advice it now says: “However, avoid using deceptive authorship information. Fabricating creator profiles (such as by using AI-generated headshots, made-up names, or false credentials to make content appear as if it was written by human experts) is a form of deception. Any form of deception makes a page untrustworthy to both users and our automated quality systems, and is a signal of a low-quality page.”

Both passages come from Google's Search Quality Rater Guidelines. The September 2025 edition we downloaded tells raters, in section 4.5.3, to rate any page using deception “Lowest”, and gives as examples AI-generated content “with made up ‘author’ profiles” and creator profiles that falsely claim credentials. Its section 3.2, on the quality of the main content, names effort, originality and talent or skill as the main measures, with accuracy added for informational pages. The difference is the audience. The rater guidelines are instructions to contractors whose ratings, Google says, do not directly affect ranking. The people-first page is advice to site owners, and it now links deception to Google's “automated quality systems”.

What it means for bylined and contributed content

For companies that publish articles under an executive's name, or place contributed pieces in trade outlets, the authorship passage is the one to read closely. Google's examples are made-up names, AI-generated headshots and false credentials. A ghostwritten article under a real executive's name, with an accurate description of who they are, is not what the examples describe. A pen name with an invented biography, a stock or AI-generated headshot on a real person's profile, or a title the person does not hold, is. How bylines work as a credibility signal more generally is set out in do author bylines affect credibility signals.

The effort definition matters for the volume end of content work. A batch of articles drafted by AI and lightly edited is exactly the example Google now gives, on its main guidance page, of little to no effort. That does not make it a spam violation; the scaled content abuse policy is about the purpose of producing many pages, as set out in what scaled content abuse is and does Google penalise AI-generated coverage. It does tell a writer how Google's raters are told to see such work.

The fact-check instruction applies to the parts readers rarely look at. If AI writes a page's title, meta description, schema markup or image alt text, Google now says a person should check those too, because they can appear in search results under the company's name.

Methodology and limitations

This is an original comparison. On 5 October 2026 we downloaded the current English versions of Google Search Central's “Google Search's guidance on using generative AI content on your website” and “Creating helpful, reliable, people-first content”, both showing “Last updated 2026-10-01 UTC”, and the documentation updates page. For the earlier versions we used copies scraped on 8 June 2026 and kept in a public GitHub repository of Google documentation, the same June baseline Search Engine Journal used. We reduced both versions to plain text, removed formatting and page furniture, split them into sentences and compared them sentence by sentence and word by word. Word counts are of the page text, including its title, and exclude navigation and the feedback widget. Three limits apply. First, the June copies are a third party's scrape, not Google's archive; the Internet Archive was unavailable when we tried it. Second, a comparison against June cannot show whether an edit was made on 1 October or at some point after 8 June; Google's last-updated date and Search Engine Journal's reporting both point to 1 October. Third, the rater guidelines quoted are the edition dated 11 September 2025, downloaded from Google on 5 October 2026. What the changes mean for ranking is not something documentation can show, and we make no claim about it.

Conclusion

Google changed two pages of content guidance on 1 October and logged one. The logged change is small and practical: check AI-written text, and the metadata around it, by hand before publishing. The unlogged change is larger. It brings the rater guidelines' definitions of effort, originality, skill and accuracy onto the page Google points site owners to first, and it calls invented author profiles deception that its automated systems treat as a sign of low quality. Neither change sets a new rule. Both state, in guidance for site owners, positions that until now sat mainly in instructions to raters. For anyone publishing under a name, the line is clear: a real person with accurate credentials is fine, whoever typed the draft, and an invented one is not.

Frequently Asked Questions

Sources

1. documentation Google Search Central, Google Search's guidance on using generative AI content on your website, last updated 1 October 2026. Read and compared 5 October 2026.
2. documentation Google Search Central, Creating helpful, reliable, people-first content, last updated 1 October 2026. Read and compared 5 October 2026.
3. documentation Google Search Central, Latest documentation updates. One entry for 1 October 2026. Read 5 October 2026.
4. primary data June 2026 copies of both pages, scraped 8 June 2026, bsisduck/google-search-ads-analytics-docs on GitHub. Third-party archive; used as the comparison baseline.
5. guideline Google, Search Quality Rater Guidelines, 11 September 2025 edition, section 4.5.3 on deception and the sections on effort, originality and skill. Downloaded 5 October 2026.
6. trade press Matt G. Southern, Search Engine Journal, Google Tells Sites To Fact-Check AI Content Before Publishing, 1 October 2026, and Google Adds Fake Author Warning To Helpful Content Guidance, 2 October 2026. Read 5 October 2026.