AI detectors and SEO: what to focus on instead
Short answer No. Google ranks pages on quality, not on how they were made, and doesn't publish any use of third-party AI detector scores. The detectors are also unreliable: OpenAI pulled its own for low accuracy. Spend the time on what Google names: first-hand experience, facts a person checked, links from real sites, and pages that answer the customer.
Want help applying this to your business? Talk it through on a free 30-minute call.
Book a callDo AI detector scores affect SEO?
No. A detector score is a guess from a third-party tool. Google has never said it uses one, and its stated rule judges the page, not the writer:
"Our focus on the quality of content, rather than how content is produced, is a useful guide that has helped us deliver reliable, high quality results to users for years." (Google Search Central Blog, February 8, 2023)
What Google does act on is content made mainly to rank. Its helpful-content page calls that "search engine-first content made primarily to gain search engine rankings" (Google Search Central). A page can score "100% human" and still be exactly that. For the full policy, see Does Google penalize AI content?
| You're tempted to... | Better use of the hour |
|---|---|
| Rewrite a page until a detector says "human" | Add one fact only you know: a price range, a photo, a step from your own jobs |
| Buy a "humanizer" tool | Have a person fact-check every number and rule |
| Run every old page through a detector | Open Search Console and fix pages that get seen but not clicked |
| Worry a page "sounds like AI" | Ask whether a customer would finish it with their question answered |
How accurate are AI content detectors?
Not accurate enough to steer your website by. Three public records:
- OpenAI shut down its own detector. In its tests it caught 26% of AI-written text and wrongly flagged human text 9% of the time. On July 20, 2023, OpenAI withdrew it "due to its low rate of accuracy" (OpenAI).
- The FTC acted on a detector's accuracy claim. Workado advertised its AI Content Detector as "98 percent" accurate; the FTC said "independent testing showed the accuracy rate on general-purpose content was just 53 percent" (FTC, April 2025). The August 2025 final order bars effectiveness claims without "competent and reliable evidence" (FTC).
- Plain writing gets flagged. Stanford researchers ran seven detectors on 91 essays by non-native English speakers; on average 61% were labeled AI-written, while essays by US eighth graders were judged almost perfectly (Liang et al., Patterns, 2023).
The last point matters for a service business. Clear, simple sentences, the kind that work best on a phone, are what the detectors in that study tended to flag as "AI." Making a page wordier to fool a tool makes it worse for customers.
What does Google look at instead?
Google says its systems look for signs of "expertise, experience, authoritativeness, and trustworthiness," and that "one of the factors used to determine quality is understanding if other prominent websites link or refer to the content" (Google: How Search works). That gives an owner four jobs.
1. Show you've done the work. Google's own self-check asks:
"Does your content clearly demonstrate first-hand expertise and a depth of knowledge (for example, expertise that comes from having actually used a product or service, or visiting a place)?" (Google Search Central)
A plumber's page that says which pipe fails first in 1960s homes in your area passes. A page any company could publish doesn't, whoever typed it. More on this in E-E-A-T for a small business.
2. Check every fact. Google's AI guidance says: "It is critical to manually factcheck and review all AI-generated content for accuracy and trustworthiness before publishing" (Google Search Central). A wrong price or rule costs you a customer; a detector score costs you nothing.
3. Earn links from real sites. Links (other websites pointing to yours) are a named quality signal. For a local business they come from suppliers, associations, sponsorships and local news. See how to get backlinks for a local business.
4. Don't mass-produce. Google's warning signs include "Are you using extensive automation to produce content on many topics?" (Google Search Central). Ten pages about services you actually sell beat a hundred thin ones.
Should I tell readers I used AI?
Only where they'd wonder. Google says disclosures "are useful for content where someone might think 'How was this created?'" (Google Search Central). A how-to page on your services usually doesn't need a label. A real name and role on the page helps more: Google suggests "accurate author bylines when readers would reasonably expect it" (Google Search Central Blog).
How do I know if my content works?
Read results, not scores. In Google Search Console, check each page's impressions (times it was shown), clicks and the searches behind them. Then check whether it brings calls, forms or bookings. A page that ranks and brings jobs is working, whatever a detector says about it.
Example
A landscaper's AI-drafted pages
The owner drafts five service pages with AI, adds her own prices and before-and-after photos, and fixes two wrong plant names. A detector says "80% AI." Google sees five specific, accurate pages. Nothing to change.
A cleaning company that "humanized" its blog
The owner pays $30 a month for a tool that rewrites posts until detectors pass. The posts get longer and vaguer; clicks don't move. Better use of the $30 and the hours: one new page answering the question customers ask on every call.
A contractor accused by a client
A client's detector flags a page the owner wrote himself. That's a false positive, which OpenAI's and Stanford's numbers show is common. His best proof is the page itself: job photos, local permit details, his name.
FAQ
Does Google use an AI detector on websites?
Google has published no such step in ranking. Its spam systems target pages made mainly to manipulate rankings, however they were made (Google Search Central Blog).
Is a free AI detector good enough to check a writer's work?
It's a weak signal at best, given the error rates above. Check the work itself: are the facts right, is anything only your business would know, would a customer act on it?
Do "AI humanizer" tools help SEO?
No ranking benefit is documented. They change wording, not facts or experience, which is what Google says it rewards.
Can a human-written page be hurt like AI spam?
Yes. Google's warning signs, such as content "primarily made to attract visits from search engines," apply to any page (Google Search Central).
Different for you?
- Worried about a penalty for AI pages: Does Google penalize AI content?
- Your pages lack trust signals: E-E-A-T for a small business
- Few sites link to you: Backlinks for a local business
- You want a team to write and maintain pages: SEO agent