You usually cannot tell a fake review by reading it, so stop trying to. The US consumer regulator says as much in its own advice. What you can check in a minute is everything around the words: when the reviews arrived, who wrote them, whether they are about this product, and whether anyone else agrees. Those checks work the same whether a fake was typed by a person or a machine.
- Fake reviews are now specifically illegal in the US and the UK, including ones written by AI and ones bought from staff or customers for a set rating.
- Reading harder does not work. Polished text is not proof of a fake, and detector tools are not reliable enough to tell AI-written reviews from human ones.
- Check the pattern, not the prose: a burst of reviews in a few days, reviewers with one review ever, praise for a different product.
- Read the middle ratings, and look for labels saying a reviewer was given something.
- Platforms say they remove millions of fakes a year. Those are their own counts of what they caught, not of what got through.
What the law now bans
In the US, the Federal Trade Commission’s rule on consumer reviews and testimonials went into effect on 21 October 2024.US Federal Trade Commission, The Consumer Reviews and Testimonials Rule: Questions and Answers, read at source 18 Sep 2026: “The Commission’s Rule on the Use of Consumer Reviews and Testimonials, which went into effect on October 21, 2024, addresses deceptive and unfair conduct involving consumer reviews and testimonials.” It covers reviews that pretend to come from someone who does not exist, and the FTC names AI as one way that happens. It also bans paying for reviews on condition they say something positive or negative, undisclosed reviews by a company’s own officers and managers, company-run review sites posing as independent, and threats used to get bad reviews taken down.US Federal Trade Commission, Federal Trade Commission Announces Final Rule Banning Fake Reviews and Testimonials, press release of 14 August 2024, read at source 18 Sep 2026: the rule “addresses reviews and testimonials that misrepresent that they are by someone who does not exist, such as AI-generated fake reviews, or who did not have actual experience with the business or its products or services”.
In the UK, fake reviews became a banned practice under the Digital Markets, Competition and Consumers Act 2024, with the new regime announced as in force in April 2025.Competition and Markets Authority, CMA to boost consumer and business confidence as new consumer protection regime comes into force, published 7 April 2025, read at source 18 Sep 2026: “The DMCCA includes an explicit ban on the posting and commissioning of fake reviews – which has been added to the banned practices list.” The regulator’s guidance defines a fake review in one line: “A fake review is a consumer review that purports to be, but is not, based on a person’s genuine experience.” Its examples of what is banned include buying reviews that look like the work of individual customers “but have in fact been generated by software applications (such as bots)”, and hiding the fact that a reviewer was paid or given something.Competition and Markets Authority, Fake reviews (CMA208), guidance dated 4 April 2025, read at source 18 Sep 2026.
The UK rules also reach how reviews are shown, not just how they are written: hiding negative reviews, or star ratings that paint an inaccurate picture. In March 2026 the CMA opened investigations into five businesses over practices of this kind, and said it had reached no conclusions.Competition and Markets Authority, Fake and misleading reviews: 5 businesses under CMA investigation, published 27 March 2026, read at source 18 Sep 2026: the banned practices cover “how reviews are handled – for example, if negative reviews are hidden, or if star ratings present an inaccurate picture”.
The law is aimed at businesses, not at you. It is still worth reading as a shopper, because a list of what is banned is a list of the tricks people were using.
What you cannot tell by reading
The FTC’s advice to shoppers is unusually frank about this: “Don’t assume that, just by looking, you can spot the difference between a real review and a fake one.” It adds that fakes are not always glowing, and not always positive: “some fake positive reviews give less than the highest possible rating in order to seem more believable”, and some businesses post fake negative reviews about competitors.US Federal Trade Commission, How To Evaluate Online Reviews, consumer advice dated October 2022, read at source 18 Sep 2026.
The UK regulator’s tips lean the other way on two points. It asks shoppers to be alert to AI-written reviews: “If a review feels a bit too slick, reads like it’s been perfectly crafted, or uses very similar wording to others, it may not reflect a real customer’s experience.” And it treats middling ratings as the safer ones: “It’s unlikely that a 3 or 4-star review is fake, so they can be particularly useful.”Competition and Markets Authority, Fake and misleading reviews: 5 businesses under CMA investigation, Top Tips section, published 27 March 2026, read at source 18 Sep 2026.
The two pieces of advice can both hold. Middling reviews may well be faked less often, and a careful faker would write exactly that kind. Very similar wording across many reviews is a pattern you can see; “too slick” on one review is a hunch. A real customer can also tidy a review with an AI tool, and the words then read as machine-made while the experience behind them is genuine. The US rule turns on whether the reviewer exists and had the experience, not on what typed the sentences.Reasoning, September 2026 — from the rule text, 16 CFR Part 465, eCFR, read at source 18 Sep 2026, which bans reviews that misrepresent “That the reviewer or testimonialist exists”, that they had experience with the product, or what that experience was.
Detector software does not rescue this. This site’s guide to spotting AI writing covers the research: detectors misfire on text they were not trained on, and they wrongly flag careful human writers. A review is a few sentences long, which gives any tool even less to go on.
Treat the wording of one review as the weakest evidence you have. Patterns across many reviews, and facts about the reviewer, are much harder to fake.
The one-minute check
Most of these come straight from the two regulators’ own advice. None needs a tool.
- Look at the dates. The FTC: “watch for a burst of reviews over a short period of time. That can sometimes mean the reviews are fake.”US Federal Trade Commission, How To Evaluate Online Reviews, read at source 18 Sep 2026.
- Open two or three reviewers’ profiles. “If it seems that the reviewer has created an account just to write one review for one product, that review may be fake.”US Federal Trade Commission, How To Evaluate Online Reviews, read at source 18 Sep 2026.
- Check the reviews are about this product. The UK guidance names “catalogue abuse”, where reviews of a different product are shown as if they were for the one you are looking at. Praise that mentions the wrong size, colour or model is the tell.Competition and Markets Authority, Fake reviews (CMA208), 4 April 2025, read at source 18 Sep 2026: “presenting reviews of a different product as relating to the product a consumer is considering”.
- Actually read them. The CMA: “You’ll be surprised at how many reviews sound dubious, overly vague or even totally unrelated to the item they’re supposedly endorsing.” A review that names a specific use, a flaw or a delivery problem has more to lose from being false.Competition and Markets Authority, Top Tips, published 27 March 2026, read at source 18 Sep 2026.
- Look for incentive labels. Some sites mark reviewers who were given a free product. The FTC leaves the weighing to you; under UK law, a site publishing incentivised reviews must set them apart from the rest.US Federal Trade Commission, How To Evaluate Online Reviews: “How you weigh those reviews is up to you.” Competition and Markets Authority, CMA208: publishers “must distinguish the incentivised review from other reviews”. Both read at source 18 Sep 2026.
- Check one other site. Both regulators say it. A place with a glowing score on one platform and a poor one everywhere else is telling you something.US Federal Trade Commission, FTC warns businesses about fake reviews, 22 December 2025, read at source 18 Sep 2026: “Look at a variety of sources and pay attention to whether a website or its reviews are independent or sponsored.”
On Google Maps, one more signal is now visible: Google says that when it sees a sudden spike in spam reviews it pauses new reviews on the profile and shows shoppers a banner explaining why.Google, New ways we’re protecting businesses on Maps, 16 April 2026, read at source 18 Sep 2026: it will “display a notification banner to let consumers know why contributions are temporarily paused”.
Every check here is about the reviews as a group or about the people writing them. That is why it works on AI-written fakes too: a machine can write a convincing sentence far more easily than it can give an account years of history.
What the platforms say they remove
The big review sites publish their own numbers. The most recent editions found for this page, each in the platform’s own words:
- Google Maps, 2025: over 292 million policy-violating reviews blocked or removed, while more than 1 billion were published.Google, New ways we’re protecting businesses on Maps, 16 April 2026, read at source 18 Sep 2026: “Our advanced systems and expert analysts blocked or removed over 292 million policy-violating reviews, while publishing more than 1 billion helpful reviews.”
- Amazon, 2025: “hundreds of millions” of suspected fake reviews blocked. The page gives no exact figure.Amazon, How Amazon Ensures Trustworthy Reviews, read at source 18 Sep 2026: “Amazon proactively blocked hundreds of millions of suspected fake reviews from our store in 2025.”
- Trustpilot, 2024: 4.5 million fake reviews removed, 7.4% of all reviews submitted that year, 90% of them by automated systems.Trustpilot, Trust Report release, 29 May 2025, read at source 18 Sep 2026: “Trustpilot removed 4.5 million fake reviews - amounting to 7.4% of the total number of reviews submitted in 2024.”
- Tripadvisor, 2024: 2.7 million fraudulent reviews kept from travellers, of which the largest share (54%) was businesses and their staff boosting themselves, and 214,000 reviews it identified as AI-generated removed.Tripadvisor, 2025 Transparency Report release, 18 March 2025, read at source 18 Sep 2026: “214,000 AI-generated reviews were flagged and removed in 2024 to prevent misleading content.”
Read these with care. Each company counts something different — “policy-violating”, “suspected fake”, “fake”, “fraudulent” — and two of the four describe 2024, not 2025, so they cannot be ranked against each other. Every figure counts what was caught, not what got through, and none of the releases describes an independent audit of its numbers. Tripadvisor’s release does not say how it decided a review was AI-written or how often that call is wrong. The questions on reading a statistic apply to all four.Reasoning, September 2026 — comparison of the four releases cited above as read at source on 18 Sep 2026.
A big removal number shows a platform is working at it. It does not tell you the page in front of you is clean. The checks above still apply.
If you spot one
Report it to the site it is on. In the US the FTC also wants to hear about it: use ReportFraud.ftc.gov, pick the path for the product or service (or “Other”), and write “fake review” in the comments field.US Federal Trade Commission, FTC warns businesses about fake reviews, 22 December 2025, read at source 18 Sep 2026: “Be sure to tell the FTC about the fake reviews you spot at ReportFraud.ftc.gov.”
Fake reviews are one part of a larger problem this site covers in pages written for machines: shops, pages and audiences built to look busy for software rather than useful to people.
What this page could not verify
- How many reviews online are fake. The platform figures above count removals, not what remains, and this page found no independent measure it could check.
- Whether Trustpilot or Tripadvisor have published newer reports. The editions cited describe 2024 and were the latest found on 18 Sep 2026.
- How platforms identify AI-written reviews, and how often they are wrong. None of the releases read for this page says.
- The UK reporting route for shoppers. Not checked for this page; report to the platform in the first instance.
US Federal Trade Commission, Final Rule Banning Fake Reviews and Testimonials (press release, August 2024) · 16 CFR Part 465 (eCFR) · Rule Questions and Answers · How To Evaluate Online Reviews · FTC warns businesses about fake reviews (December 2025). Competition and Markets Authority, new regime in force (April 2025) · Fake reviews guidance CMA208 (April 2025) · 5 businesses under investigation, with Top Tips (March 2026). Google, New ways we’re protecting businesses on Maps (April 2026) · Amazon, How Amazon Ensures Trustworthy Reviews · Trustpilot, Trust Report release (May 2025) · Tripadvisor, 2025 Transparency Report release (March 2025). All read at source on 18 September 2026.