Should you let AI write your website content? What Google actually penalises, what it does not, and how to use it without wrecking your rankings

Your designer says use ChatGPT for everything. Your agency says Google will penalise you. Neither is telling you the whole story.

Share

What Google actually says

Google does not penalise content because AI wrote it. It rewards content that is helpful, original and trustworthy, and it penalises content produced at scale primarily to manipulate rankings, whoever or whatever produced it. That is the position in Google’s own guidance on AI-generated content and in the spam policies, which now name “scaled content abuse” directly. The question to ask about any page is not “did AI write this” but “would anyone be glad they found this, and does it say anything the top ten results do not”.

The real risk, which is not what you were told

The fear people sell is detection: Google spots the robot and punishes you. The actual failure is duller. AI, asked to write “a page about emergency plumbing in Leeds”, writes the average of every emergency plumbing page it has read. It is grammatical, plausible, and interchangeable with your competitors’ AI pages. Google has nothing to rank it above them for. AI engines have nothing to cite it for. The page does not get penalised; it gets ignored, which costs you the same money more quietly.

Multiply that by two hundred pages, one per town or product, and you cross from “ignored” into “scaled content abuse”, where the whole site can be actioned.

Safe, risky, and reckless

UseVerdictWhy
Turning your own notes, a call recording or a voice memo into a first draftSafeThe substance is yours; the tool is a typist
Rewriting an existing page for clarity, shorter sentences, plain languageSafeImproves a page that already has your knowledge in it
Drafting FAQs from questions customers actually asked youSafeReal questions, your answers, tidied
Translating pages you wrote, checked by a native speakerSafeSame content, more people
Generating a whole page from a keyword and publishing it after a skimRiskyGeneric, unverifiable, indistinguishable from competitors
Letting AI invent statistics, quotes, case studies or reviewsRecklessFabrication; a trust and sometimes a legal problem
Hundreds of location or product pages generated from a templateRecklessThis is the definition of scaled content abuse

A worked example

Composite example: details changed, numbers typical of what I see rather than one named client.

A home services company launches 140 pages, one for every town in the county, generated from a template in an afternoon: “Looking for a reliable electrician in [Town]? Our experienced team…” Two months later, the site’s impressions in Search Console halve and a “scaled content abuse” manual action appears. The pages that had ranked for years, the genuine service pages, are dragged down with them. Recovery: delete the 140 pages, keep five real area pages with photos of jobs done in each area, named team members who cover it, and real travel times, then file a reconsideration request. The manual action lifts after five weeks. The five real pages outrank the old 140 combined within three months, because they say something true.

The workflow that keeps you safe

  1. Start with what only you know. Your prices, your process, what goes wrong, what you refuse to do, a real example, your photo. Write it as notes; badly is fine.
  2. Let AI draft from those notes only. Give it your notes and say “use nothing that is not here”. If it adds a claim you did not give it, delete the claim.
  3. Read every sentence and ask “is this true, and is it ours”. Cut everything that could appear unchanged on a competitor’s site.
  4. Add the things AI cannot. A number, a screenshot, a quote from a real customer with permission, an opinion you would defend on a call.
  5. Publish fewer pages. One excellent page per service beats twenty adequate ones, for rankings, for AI citations, and for the person reading it.

What this means for you

  • Local service business: use AI to tidy your service pages and FAQs from your own knowledge. Never generate town pages. Your photos and your reviews are worth more than any text.
  • E-commerce: AI is fine for first-draft product descriptions from real specifications, checked for accuracy. It is not fine for inventing benefits or for a thousand identical category intros.
  • B2B or SaaS: your edge is expertise. Record a subject-matter expert talking for fifteen minutes and draft from the transcript. That produces pages the generic tools cannot, and they are the ones that get cited.

If you already have a site full of generated pages and want to know whether it is in the “ignored” zone or the “actioned” zone, send me the URL. The Search Console pattern tells you within minutes, and the fix is usually smaller than you fear.

Quick answers

Can Google detect AI-written content?

Google has said it does not use AI detection as a ranking signal; it evaluates whether content is helpful, original and trustworthy. Detection tools you can buy are unreliable in both directions. Plan on being judged by usefulness, not by authorship.

Is it safe to use ChatGPT for blog posts?

For a first draft from your own outline and notes, yes. For a finished post generated from a title, no; it will read like every other AI post on the topic, and nothing generic gets ranked or cited any more. Add what only you know before publishing.

What is "scaled content abuse"?

Google's name for producing many pages primarily to rank rather than to help, whether by AI, by scraping or by cheap outsourced writing. It is a spam policy violation and can lead to a manual action against the whole site.

Do I need to disclose AI use on my website?

Google does not require it. Some industries and countries have rules about AI-generated material in specific contexts, and your own readers may care. A short note on how content is produced, where relevant, costs nothing and builds trust.

Sources, and what each one says

Summarised here so you do not have to leave. Links open in a new tab if you want to check for yourself.

  1. Google Search Central: Google Search's guidance about AI-generated content developers.google.com

    Google's 2023 statement: AI-generated content is not against its guidelines; quality and originality are what count.

  2. Google Search Essentials: spam policies (scaled content abuse) developers.google.com

    What Google counts as spam: scaled content abuse, link spam, doorway pages, hidden text and more.

  3. Google: creating helpful, reliable, people-first content developers.google.com

    The self-assessment questions Google suggests for judging whether a page is people-first.

Share
If it would help

Want me to look at your site?

Send the URL. I read it myself and reply with the three things I would fix first. No deck, no drip sequence, no charge.

Free, and I reply personally. Privacy