Google's helpful content page now defines main content, effort and fake author profiles. Here is what it says about AI text and what to fix on your blog.
If your business publishes blog posts written mostly by AI, or uses a made-up author name to make them look more credible, Google's latest guidance is aimed squarely at you. Google's page on creating helpful, reliable, people-first content now spells out what it means by main content, and how it judges the work behind it.
SEO consultant Marie Haynes flagged the edit in a video on 2 October 2026, and PPC Land covered it on 7 October. The timing matters: the September 2026 spam update only finished rolling out on 8 October, after about 13 days. Here is what the page says, what is still unclear, and what to check on your own site this week.
What Google now says counts as main content
Google defines main content as "any part of the web page that directly helps the page achieve its purpose." That is wider than most people assume. According to PPC Land's reading of the update, it covers the primary text and media, but also interactive tools such as calculators, user comments, tabbed or expanded sections, and page titles and headings.
So a service page with a pricing calculator is judged on the calculator too, not just the paragraphs around it. A thin template with a few swapped city names does not become useful because the page looks busy.
The four things Google looks at
The page uses four attributes to judge main content. Google's own wording is on its help page, and it is worth reading in full.
- Effort: how much human work went into the content. Google's example of low effort is "using generative AI to produce large amounts of text without manual oversight or curation."
- Originality: whether the page offers "unique, original information or perspectives that aren't already available on other websites."
- Talent or skill: whether the content shows the skill or expertise needed to give visitors a satisfying experience.
- Accuracy: for topics that affect health, money or safety, the content must be "highly accurate and consistent with established expert consensus."
One line is easy to miss: attribution or credit to other sources "doesn't replace the need for original effort." Rewording three competitor articles and linking to them is still low effort.
AI is not banned, but unsupervised AI is the problem
Read the effort line again. The problem Google names is generating large amounts of text without manual oversight or curation. It does not say that any use of AI is penalised. Haynes read the new wording as accepting AI help when real human work goes alongside it, though that is her interpretation, not a statement from Google.
The page also asks whether the use of automation, including AI generation, is "self-evident to visitors through disclosures or in other ways." If readers would reasonably expect to know how a piece was made, tell them.
In practice, the line between a tool and a content mill is whether a person with real knowledge of your business shaped, checked and added to the draft. A founder's own numbers, a customer story, a photo you took yourself: these are things a model cannot invent for you.
Fake author profiles are now called deception
The most pointed addition is about authors. Google says fabricating creator profiles, such as using AI-generated headshots, made-up names or false credentials, is deception, and that it makes a page untrustworthy.
Many small sites have a fake "Dr. Sarah Mitchell, 15 years of experience" at the bottom of their health or finance posts. If that describes you, remove it. A real person with a real, modest bio is safer than an invented expert.
Haynes also notes that bylines are not a ranking factor, though Google encourages accurate authorship where readers would expect it. So the aim is not to add authors everywhere. It is to make sure any author you show actually exists.
What we do not know yet
It is worth being straight about the limits here. PPC Land's report relies on Haynes reading the new wording aloud, and the exact edit time is not stated. Nothing published says rankings have changed because of the edit, and nothing describes a penalty or how Google would detect a fabricated profile.
Haynes speculates the edit could precede a core update, but she hedges that one may not come. Treat that as a guess. What is confirmed is the text on Google's own page, and that text is a clear statement of what the company wants to reward.
There is also some context from the spam update. Haynes said she has seen requests for help with manual actions go from almost none to several a week lately, and some site owners report thin content actions even after removing thousands of URLs. That is anecdotal, but it fits the same direction.
A 30 minute content check you can run this week
- List your 20 most visited blog posts and mark which ones a person with real expertise has actually reviewed.
- Remove or replace any author name, headshot or credential that is not real. Use a genuine staff member, or publish under the company name.
- Look for posts that only restate what the top five search results already say. Add something they lack: your data, a worked example, a screenshot, a price range you actually charge.
- Check anything that touches health, money, legal or safety topics against an authoritative source, and cite it.
- Add a short disclosure where AI helped and readers would expect to know.
- Merge or delete near-duplicate pages that exist only to catch a keyword variation.
How to use AI without writing low effort content
Use it for the jobs that sit around the writing: outlining, summarising your own interview notes, tightening a draft, checking for gaps. Keep the facts, opinions and examples human. If you cannot say what a post adds that the search results do not already have, it is not ready.
Our content marketing and blog writing teams work this way, and our AI content service is built around human review rather than bulk output. If you are unsure where your site stands, a digital marketing audit will show which pages are likely to be seen as thin.
Key takeaways
- Google's helpful content page now defines main content and judges it on effort, originality, talent or skill, and accuracy.
- Mass AI text with no manual oversight is the named example of low effort. Supervised AI use is not called out as a problem.
- Made-up author names, AI headshots and false credentials are described as deception.
- No ranking change or penalty has been confirmed from this edit, so do not panic, but do not wait either.
- Start with your top 20 posts: real authors, original details, and human review.
If you would like a second pair of eyes on your content, you are welcome to get a free quote and we will tell you honestly what we would change first.
John Doe
Supreb Information. Thanks
15 July 2014 View replies . 1