all case notes

Content, social and email

Will Google penalise my business blog if I use AI to write it?

Google does not ban AI writing. It does object to mass-produced pages with no value. Here is where the line sits, and a simple review routine for AI drafts.

By Shany BouwerPublished 13 November 20265 min read

A vintage typewriter beside a laptop on a wooden desk, seen from above

A question I hear from owners, usually in a lowered voice, as if asking about something slightly illegal: "Can I use AI to write my blog? Will Google punish me?"

Short answer: Google does not punish a page because a machine helped write it. It does object to a particular way of using machines. The difference is worth knowing, because the wrong guess costs you either a lot of time or a lot of trust.

Does Google penalise AI-written content?

Not simply for being AI-written. Google's guidance on using generative AI content says generative AI can be useful when researching a topic and for adding structure to original content. The warning comes next: using AI tools to generate many pages without adding value for users may violate Google's spam policy on scaled content abuse.

So the question Google cares about is not "who typed it". It is "did this page add something for the reader, or was it one of a thousand copies made to catch searches?"

What is scaled content abuse, in plain words?

Scaled content abuse is Google's name for producing lots of pages mainly to manipulate search results, not to help anyone. Google's spam policies describe it, and the examples are not limited to AI, and the definition applies "no matter how it's created".

Google's page on people-first content puts the good version simply: content created primarily for people, and not to manipulate search rankings.

A business owner writing eight honest posts a month with help from a tool is nowhere near that. A site that publishes five hundred near-identical "best plumber in every suburb" pages overnight is.

What is my take on AI and business blogs?

My view: use AI as a fast, slightly overconfident assistant, never as the author. It is good at structure, at first drafts of boring parts and at spotting what you forgot to say. It is bad at knowing your business, and it will state wrong things in a calm voice.

I use AI tools, mainly Claude, to do the repetitive parts of my work. I design the process, write the rules and sign off every change. Nothing reaches a customer without my approval. That is not modesty, it is the only arrangement I trust. This is my opinion and my own way of working, not a Google requirement, so adapt it to your risk appetite.

Why can an AI draft be risky even when it sounds fine?

Google says it plainly: generative models do not retrieve facts, they predict a likely sequence of words from training data, so their output can contain inaccuracies, often called hallucinations. The same page of guidance says it is critical to manually fact check and review AI-generated content for accuracy and trustworthiness before publishing, including things like title elements and meta descriptions.

I have a small, real reason to take that seriously. In my own work on a monthly supplier catalogue, a tool quietly returned 20 of 30 products and gave no warning. Nothing looked broken. A third of the catalogue would have gone unchecked if I had not counted. An AI draft fails the same way: smooth on the surface, wrong in the middle, and silent about it. The lesson I took is to check the count, not the confidence.

How do I review an AI draft before it goes live?

Here is a routine you can run this week. It takes 20 to 30 minutes per post.

  1. Check every fact and number. If you cannot point to where it came from, delete it. Names, prices, dates, laws and "studies show" are the usual offenders.
  2. Check every link. Open each one. Make sure it loads and says what the sentence claims.
  3. Add what only you know. A real customer question, a photo, a price you actually charge, a mistake you have seen. This is what stops the post being one of a thousand copies.
  4. Read it aloud. If it sounds like a brochure written by nobody, rewrite it until it sounds like you.
  5. Check the title and description. Google's guidance says this review applies to them as well. If you want help there, see how to write a page title.
  6. Publish fewer, better posts. Eight posts you stand behind beat thirty you skimmed.

Should I tell readers I used AI?

Google's guidance suggests that giving background on how automation was used can be helpful for readers, and it is worth reading that section yourself. Whether you add a note is up to you. My habit is simple: if a reader would feel misled to find out, say it. If the tool only tidied my sentences, I do not think a note is needed. That is a judgment call, not a rule.

What should I do next?

Pick one existing post and run the six steps. If you find something wrong, you have learned how much to trust your draft process. If you would like help setting up a sensible workflow so that AI does the boring parts and you keep the final say, book a call. The checklist works without me, though.

For how assistants such as ChatGPT decide whether to mention a business, there is a separate post: does my business show up in ChatGPT.

Frequently asked questions

Can Google detect AI-written text?

I cannot tell you how Google's systems work internally, and Google's guidance does not frame the issue as detection. It focuses on whether content is helpful and whether it meets the spam policies. Write for the reader and the question goes away.

Is it safe to publish AI content as long as I edit it?

Editing is the right habit, but the test is value and accuracy. Fact checking, adding your own knowledge and publishing in sensible volume are what keep you on the right side of the guidance.

How many posts a month is too many?

There is no number in Google's guidance. The warning is about producing many pages without adding value. I would rather publish eight careful posts than thirty rushed ones, but that is my preference, not a threshold.

Do I need to disclose that AI helped?

Google does not require a label in the guidance I read. Be honest if your readers would reasonably want to know, and check any rules that apply in your industry or country.

Got a problem nobody's cracked yet?

Tell me what is broken or what you are building. I reply within one working day.