Google and AI drafts: what the guidance says and what editors read into it
Teams keep asking whether AI-written pages get demoted. The published guidance is shorter and more boring than the rumours, and it is worth reading line by line.
Our reading
The method is not the problem; the volume and the intent are
Google's wording separates helpful pages from pages made mainly to rank. A domain trust rank checker will not make a page helpful, but it helps you avoid citing sites that carry no trust at all.
- AI use is not against Google's rules by itself (guidance of February 2023)
- Scaled content abuse is a named spam policy since March 2024
- Review and evidence of care are what you can control
Myth against reality
| What people repeat | What is on the record |
|---|---|
| Google detects AI text and demotes it | Google has said it rewards helpful content however it is produced, and it names spam behaviour, not tools |
| Adding a disclosure label fixes it | A label is a reader courtesy; it does not change whether the page is helpful or spammy |
| A human byline makes any draft safe | Quality guidance asks about experience, expertise and trust; a name without substance does not supply them |
| Links from big sites cancel thin content | Links and page quality are separate signals; see our links and AI pages note |
| Only the March 2024 update matters | Core updates keep rolling, and the August 2024 core update again reshuffled many sites |
Dates worth knowing
Helpful content update
Google launched a system aimed at content written for search engines first and people second.
Guidance on AI-generated content
Google stated that appropriate AI use is not against its guidelines, while automation used mainly to manipulate rankings is spam.
Core update and new spam policies
The helpful content ideas were folded into core ranking, and scaled content abuse became a named policy, with enforcement from May 2024.
Another core update
A further broad update that many publishers noticed; no new AI rule came with it.
Where a trust score helps and where it does not
Spam policies talk about scale and intent, and your own site cannot change a rule. It can only show care, in a form that someone outside your team could inspect later if they ever asked. That is cheap to build.
Checking outbound sources is a small piece of that care: a bulk lookup of a draft's domains costs $0.0001. A reviewer who sees three zero-score domains in a 900-word piece will stop and read, which is exactly the point. Short loop.
Be careful with the opposite error. A high DR does not make a claim true, and a brand-new specialist site, for instance a university lab page, may score low and still be the best source available. The number sorts the queue, and judgment closes it. Always.
Before you publish at volume, read our pre-publish checklist. Then see the bulk workflow example, which shows what a list of 31 domains looks like once it has been scored and sorted from the lowest number up. Review, record, repeat.
Evidence of care you can keep on file
- The brief or prompt that produced the draft
- The name of the person who edited and approved it
- A list of cited domains with their scores and the check date
- Notes on claims that were changed or removed
- The date of the last review of the page