Are AI-Generated Google Reviews Allowed? What Google and the FTC Actually Say

The short answer: a fake review is banned whether a person or an AI wrote it, but a real customer using a writing aid to describe their own real visit, then checking it and posting it themselves, is not a fake review. Google and the FTC both judge a review by whether it reflects a genuine experience of a real customer. They do not judge it by which tool helped put the words together. The line is not "AI or no AI". The line is "real customer, real experience, their own words and their own decision to post" versus anything else.

That distinction sounds simple, and it is, but a lot of businesses get it wrong in both directions. Some avoid any review help at all because they assume AI means trouble. Others buy tools that write reviews for customers who never visited, or post them from staff phones, and assume that because a machine did it the rules are fuzzy. Neither is right. This guide walks through what the rules actually say, where the line sits in practice, and the specific patterns that turn a legitimate writing aid into a fake review operation.

In the United States, the key document is the FTC's Trade Regulation Rule on the Use of Consumer Reviews and Testimonials. The FTC announced it in August 2024, it was published in the Federal Register on August 22, 2024, and it took effect on October 21, 2024. The rule makes it unlawful for a business to write, create, or sell a review that misrepresents that the reviewer exists, that the reviewer actually used the product or business, or what the reviewer's experience was. In the commentary that accompanies the rule, the FTC states plainly that AI-generated reviews are covered. It singled out AI because these tools make it cheap to produce large numbers of realistic but fake reviews.

Notice what the rule targets. It is the misrepresentation, not the keyboard. A review that claims a visit that never happened is fake. A review that invents an experience the customer did not have is fake. A review attributed to a person who does not exist is fake. All of those are prohibited whether they were typed by an owner at midnight, written by a paid agency, or generated by a chatbot. What the rule does not do is ban a real customer from getting help with wording.

The rule also bans offering compensation or other incentives conditioned on a review expressing a particular sentiment, positive or negative, and it bans certain forms of review suppression, such as intimidation or groundless legal threats used to get a review removed. You can read the full text in the Electronic Code of Federal Regulations, 16 CFR Part 465. Penalties are civil, not criminal, and they are large. The FTC's inflation-adjusted civil penalty figure for knowing violations is up to $53,088 per violation, the 2025 amount, which the FTC kept unchanged for 2026. Because every fake review can count as a separate violation, the exposure adds up fast.

Google's Maps user contributed content policy says contributions should reflect a genuine experience at a place or business, and that fake engagement is not allowed and will be removed. Its list of fake engagement includes content that is not based on a real experience, reviews that have been paid for directly or in kind, and content posted from multiple accounts by or at the request of one person. It also tells merchants not to offer incentives such as payment, discounts, or free goods in exchange for a review, and not to discourage negative reviews or selectively solicit positive ones.

Google's policy does not mention AI writing tools by name. Its test is the same as the FTC's: is this a genuine account of a real experience? A machine-written review of a restaurant the author never visited fails that test. A customer who ate there last night, answered a few questions about the meal, and posted a review based on those answers passes it.

There is one more line in Google's policy that every business using any review tool should read twice. Merchants should not require or pressure customers to leave ratings or write reviews while on the premises, and should not request that specific content be included. That is a rule about behaviour, not software. A QR code on a table that a guest can choose to scan is an invitation. A staff member standing over a guest until the review is posted, or telling them what to write, is pressure. The tool does not change which side of that line you are on.

Put the two sets of rules together and a clear test emerges. A review is legitimate when all four of these are true:

1. The reviewer is a real customer who actually visited or used the business.
2. The content reflects that customer's own experience, including the parts they did not like.
3. The customer reads the text, can change anything, and chooses whether to post it at all.
4. The customer posts it themselves, from their own Google account, with no reward attached.

If any one of those breaks, you have a problem, regardless of how the words were produced. A spell checker, a friend helping with phrasing, or an AI drafting assistant all sit comfortably inside this test as long as the four conditions hold. Plenty of people find it hard to turn a good evening into three tidy sentences. Helping them do that is no more deceptive than handing them a pen.

The risky patterns are easy to recognise once you know what to look for. Each of these turns a writing aid into a fake review generator, and each is the kind of thing both Google and the FTC describe:

• The business writes or posts the review. If an owner, manager, or employee generates a review and posts it, or posts it from a customer's phone on their behalf, the customer is no longer the author.
• The draft invents details. If the AI adds a dish the customer did not order, a staff member they never met, or praise they never expressed, the review misrepresents their experience.
• Reviews come from people who never visited. Generating reviews for a list of contacts, friends, or a review exchange group is the textbook fake review.
• Every review sounds the same. Identical or near identical text across many reviewers is exactly the kind of unusual pattern Google says it looks for and removes.
• Rewards are attached. A discount, free dessert, or entry into a draw in exchange for a review breaks Google's policy outright, and the FTC rule bans incentives tied to positive sentiment.
• Only happy customers are asked. Screening customers first and sending only the satisfied ones to Google, often called review gating, is selective solicitation that Google prohibits.
• Volume is pushed artificially. Dozens of reviews appearing in a burst because a tool was used to blast them out looks like manipulation, because it usually is.

The gating pattern deserves extra attention because it is often built into review software by default. The flow asks the customer for a star rating first, sends four and five star customers to Google, and quietly routes everyone else to a private form. It feels like smart reputation management. In practice it is selective solicitation, which Google's policy forbids, and when it is combined with misleading claims about a business's reviews it can draw the FTC's attention too. Adding AI on top of a gated flow does not make it safer. It just makes the filtered reviews read better.

A compliant flow keeps the customer in charge at every step. In concrete terms, it looks like this. The customer chooses to start, usually by scanning a code or tapping a link. They answer a handful of short questions about their own visit. The AI drafts a review using only those answers, so if the customer said the wait was long, the draft says the wait was long. The customer reads the draft, edits or rewrites it, or throws it away. If they want to post, they open Google themselves and post from their own account. Nobody is rewarded, nobody is filtered out, and nobody is told what to say.

Two design details matter more than they look. First, a negative experience must produce an honest draft, not a softened one. A tool that turns "the food was cold" into "a lovely evening" is inventing an experience. Second, every customer must get the same route to Google, whatever they said. It is fine to also offer an unhappy customer a private way to tell the business what went wrong, as long as it sits alongside the Google option rather than replacing it.

Kaisah was built on the test above, so it is worth being specific about how it works. A customer scans the business's QR code and answers about five quick questions about their own visit. Kaisah's AI turns those answers into a draft review. The customer can edit the draft freely, and they post it themselves on Google from their own account. Kaisah never posts for anyone.

• Every customer gets the Google option. There is no star rating gate and no setting that hides Google from unhappy customers.
• Unhappy customers can also choose to send private feedback to the business, offered alongside the Google path, never instead of it.
• There are no incentives. Kaisah has no rewards, discounts, or prize mechanics tied to reviews.
• The draft follows the customer's answers, including critical ones, and the customer has the final word on every sentence.

None of that makes any business immune from its own behaviour. If staff pressure guests at the table, offer a free drink for a review, or post reviews themselves, no software can make that compliant. What a well designed tool can do is make the honest path the easy one.

AI did not create a new category of allowed or banned reviews. It made fake reviews cheaper to produce, which is why regulators now mention it by name, and it made honest reviews easier to write, which is why customers like using it. The rules still turn on the same questions they always did. Is this a real customer? Is this their real experience? Did they decide what it says and whether to post it? Did anyone get paid or filtered? If you can answer those honestly, the tool that helped with the wording is not the issue. If you cannot, no amount of careful prompting will fix it.