Can customers use AI to write Google reviews?

Google doesn't prohibit a customer from getting help with wording. It prohibits reviews that do not reflect a genuine experience at the business.

A real customer who uses a writing tool to say what actually happened is leaving a legitimate review. A review generated from nothing — no visit, no job, no real opinion — is fake engagement, whether a person wrote it or a model did.

Every hard question about AI and reviews resolves to that one test: did a real customer have this experience, and do the words say what they actually think?

Google's contributed-content policy bans what it calls fake engagement: content that does not represent a genuine experience. That covers paid reviews, reviews from people who were never customers, incentivized reviews, and posting on someone else's behalf.

Notice what's not on that list: writing assistance. Spellcheck, autocomplete, dictation, a translation app, or an AI draft — none of these make a genuine review fake. Fabrication makes a review fake.

Google also detects patterns, not just individual reviews. Bursts of similar-sounding text from similar accounts get filtered together, which is why mass-produced AI reviews tend to vanish even when each one looks plausible alone.

Nobody thinks a review is fake because the customer used spellcheck. Nobody thinks dictating a review into a phone makes it dishonest. The tool was never the issue — the source of the facts is.

Picture two reviews. In the first, a customer tells an assistant: the tech arrived at 8, found the leak under the slab, explained the options, and the price matched the quote — and the assistant shapes that into clean sentences. In the second, someone types 'write a 5-star review for a plumber' and pastes the result.

Same technology. Opposite sides of the line. The first review contains a real morning. The second contains nothing at all.

Set policy aside for a moment. A review generated from nothing has a deeper problem: it has nothing to say. It praises 'great service' and 'professional staff' because that's all a model can write without facts to work from.

Real reviews have fingerprints — the employee's name, the weird noise the AC made, the crew showing up early on a Saturday. Future customers trust those details precisely because they're hard to invent. A wall of smooth, interchangeable praise reads as fake to humans long before any filter flags it.

So even a business willing to cheat gains almost nothing. The reviews that actually win customers are the ones with texture, and texture only comes from someone who was there.

There's a version of AI help that makes reviews more honest, not less — because it captures detail that the blank box loses.

It looks like this: the customer rates the experience, taps the things that stood out, answers a couple of quick questions about what happened, and adds anything extra in their own words. The AI's only job is shaping those real answers into readable sentences. The customer reads the draft, edits anything that's off, and posts it themselves.

That's what guided reviews are. Every fact in the draft traces back to something the customer actually said. Nothing posts automatically. The customer approves every word — which is more scrutiny than most hand-typed reviews ever get.

The same line applies to owners, with higher stakes — because businesses caught manufacturing reviews face profile suspensions, not just deleted posts.

Generating reviews for your own business and pasting them in is straightforward fraud, and the FTC has fined businesses for it. Writing a review with AI and handing it to a customer to post is the same thing wearing a customer's name. Auto-posting on a customer's behalf removes the one safeguard that makes assistance legitimate: their approval.

small Talk exists because the blank box kills honest reviews, not because the world needs more synthetic ones.

The flow only works with a real customer's real answers. Low ratings get an equal path — post publicly or tell the business privately, their choice. The draft is theirs to edit. The posting is theirs to do. The AI does the one thing it should: help a person who was actually there find the words.