Fake Review Statistics: How Many Online Reviews Are Fake in 2026?

Online reviews influence billions of dollars in consumer spending every year. 89% of people use reviews when researching a product or service, and BrightLocal’s 2026 Local Consumer Review Survey puts that number at 97% for local businesses in the US.

But how many of those reviews are fake?

Key Fake Review Stats

• Between 11% and 15% of reviews are likely fake.
• Consumers correctly identified AI-generated fake reviews 53.2% of the time, barely better than a coin flip.
• Nearly 14% of 73 million Google reviews analyzed across home services, legal, and medical categories were judged extremely suspicious.
• On products actively purchasing fake reviews, 56% of reviews are fake on average, compared to just 0.6% on honest products.
• Google blocked or removed more than 292 million policy-violating reviews and 13 million fake Business Profiles in 2025.
• Trustpilot removed 7.8 million fake reviews in 2025 from a base of 361 million active reviews.
• Yelp filtered nearly 500,000 suspected AI-generated reviews in 2025 and closed more than 1.3 million user accounts for policy violations, a 138% increase from 2024.
• FTC penalties for fake review violations reach up to $53,088 per violation.
• 97% of consumers think businesses caught using fake reviews should face consequences.

How Many Online Reviews Are Fake?

There’s no single, universally accepted answer to this question. The published estimates range from around 4% to as high as 42%, and the inconsistency is mostly related to methodology.

The Published Range

At the conservative end, an academic literature review from the Center for Data Innovation (ITIF), cross-referencing transparency reports from Yelp and Trustpilot, estimated that roughly 4% of all online reviews are fake (2022). Tripadvisor self-reported a similar number, putting fake submissions at 3.6% of all reviews in 2020.

The UK government’s research uses a higher number. Using a trained classifier run against live marketplace data across consumer electronics, home and kitchen products, and sports equipment, the Department for Business and Trade found 11% to 15% of reviews were fake. That figure is the most credible platform-wide estimate available because the method, the data, and the limitations are all published.

The highest figure in wide circulation is Fakespot’s assessment that roughly 42% of about 720 million Amazon reviews were fraudulent in 2020. That number entered the legislative record through a US Senate letter to Amazon’s CEO, which is why it frequently gets misattributed as a government finding. It’s actually Fakespot’s estimate, based on a proprietary and unpublished algorithm.

Why the Estimates Disagree

The range is so wide because these studies are measuring different things. Some count suspicious review text, others count reviewers, and others count products with evidence of purchasing reviews. Those produce very different denominators from the same marketplace. A 4% prevalence rate on Yelp means something different from a 16% filtered-review rate, which means something different from 20% of products having at least one purchased fake review.

Whether the source has a commercial interest also matters. The Transparency Company sells review-verification services. Fakespot sells a browser extension that flags fake reviews. Pangram Labs sells AI-detection software. That doesn’t mean their data is wrong, but they potentially benefit from higher percentages of fake reviews.

Platform Averages Can Be Misleading

Platform-wide percentages in the low teens sound manageable. But they’re averages, and they flatten the picture badly.

The UCLA Anderson study is built on ground-truth data, meaning the researchers observed specific sellers recruiting fake reviewers inside private Facebook groups and could identify exactly which products cheated. Among those products, the median share of fake reviews was 59%. For honest products in the same network, just 0.6% of reviews came from the same pool of fake reviewers.

Platform-wide percentages can’t address these details. A marketplace might show a 12% overall fake rate, but more than half the reviews of certain products are fake. If you’re a consumer looking at a product, how do you know if less than 1% or more than 50% of the reviews are fake?

The study also measured at the reviewer level and found that roughly 7% of unique reviewers (about 27,000 out of 368,000) in a network of 25,840 Amazon products were identified as fake. That’s interesting because it points to a relatively small number of prolific accounts creating a disproportionate share of the fake content.

What Fake Reviews Cost

The $152 Billion Figure and Where It Comes From

The most widely quoted dollar figure in this space is $152 billion. You may see it attributed to the World Economic Forum, to the University of Baltimore, and to various other sources depending on the article. The actual derivation is buried in a footnoted ITIF policy report from 2022.

Here’s the math behind it. An estimated 89% of global e-commerce revenue is influenced by online reviews. In 2020, global e-commerce revenue was about $4.28 trillion, so the review-influenced share was roughly $3.8 trillion. Apply a 4% fake-review rate, and you get approximately $152 billion in consumer spending influenced by fake reviews each year.

That arithmetic is transparent and uses conservative assumptions. But the number is anchored to 2020 revenue. Global e-commerce has grown significantly since then, so the figure may actually be low.

Government and Academic Cost Estimates

The UK Department for Business and Trade estimated that fake review text on consumer products alone causes £50 million to £312 million in annual harm to UK consumers (2023). The report explicitly states that the true figure is likely higher because it excludes services and inflated star ratings.

An NBER working paper using a randomized experiment found that fake reviews cause consumers to choose lower-quality products, with estimated welfare losses of about $0.12 for every dollar spent (2023). The same paper found that a low-cost educational intervention reduced the harm.

A separate equilibrium model from the UCLA Anderson researchers put the net consumer welfare loss at about 0.6% of the median product purchase price, or roughly $0.16 per consumer per week. And the study’s most counterintuitive finding was that most of the damage comes from general mistrust of ratings rather than from direct deception. Honest sellers get hurt by skepticism that their real reviews aren’t real, even when they are.

For home services, legal, and medical categories specifically, The Transparency Company estimated $300 billion in annual US consumer harm, or about $2,385 per household per year.

The Effect on Purchase Decisions

In a UK government experiment, consumers were 3.1% more likely to buy a product carrying well-written fake reviews. That effect nearly tripled to 9.2% when the product price exceeded £80. The effect was strongest for consumer electronics and higher-priced goods, which makes sense. The more money is on the line, the more weight a review carries.

Who Is Actually Writing Fake Reviews?

Most Come From the Business Itself

Tripadvisor published a breakdown of its fraud composition in its 2025 Transparency Report. In 2024, review boosting by business owners, employees, and affiliates accounted for 54% of all fraudulent review submissions. Member fraud was just over 39%. Vandalism was 4.8%. And paid reviews from brokers were just 2.1% of the total. Around 9,000 businesses received warnings over incentivized reviews, and 360,000 removed reviews were linked to employee incentive programs.

The UK’s Competition and Markets Authority reinforced this in March 2026 when it opened five formal investigations into Autotrader, Feefo, Dignity, Just Eat, and Pasta Evangelists. Dignity is being investigated over whether staff were asked to write positive reviews about crematoria services. Pasta Evangelists is being investigated for offering customers discounts on future orders in exchange for five-star reviews on delivery apps, without disclosing the incentive. These are ordinary businesses whose review programs crossed a line.

The Broker Market

Researchers documented more than 20 private Facebook groups where Amazon sellers recruit fake reviewers. These groups averaged about 16,000 members each, with more than 500 posts per day, and the researchers estimated that reviews were solicited on behalf of as many as 4.5 million products in a single year (He, Hollenbeck and Proserpio, Marketing Science, 2022). The data was hand-collected by research assistants working inside the closed groups because the groups can’t be scraped.

The mechanism is well-documented in the UCLA Anderson working paper. A seller posts the product in a private group. A recruited reviewer buys it at full price. They leave a five-star review. Then they get reimbursed through PayPal for the purchase price plus fees, with payment conditional on the review staying up. Because the reviewer made a real purchase, the review carries the “verified purchase” badge.

That’s the mechanism that makes the verified purchase badge unreliable as an authenticity signal. The badge confirms a transaction happened, but it doesn’t confirm the review is genuine.

Amazon reported more than 10,000 fake review groups to Meta since 2020. In 2025 alone, Amazon’s legal actions led to the shutdown of more than 100 websites facilitating fake reviews, with over 40 brokers ceasing their activity.

And one additional finding from the Marketing Science study that gets almost no attention: fake reviews are mostly purchased by established listings, not new ones. Of the roughly 1,500 products observed buying reviews, only 17 did so in their first week on Amazon, and only 94 within their first month. The common advice to be suspicious of products with few reviews may be inaccurate.

Fake Negative Reviews and Extortion

In late 2025, a wave of extortion scams hit businesses on Google Maps. Google described the pattern in its November 2025 fraud advisory: attackers flood a business profile with fake one-star reviews, then contact the owner through a third-party messaging app and demand payment to stop. Google launched a dedicated extortion reporting form and, in its April 2026 update, said its systems had been upgraded to detect and block these attacks before the reviews go live. Yelp’s 2025 Trust and Safety Report noted that its team prevented review extortion scams that had impacted businesses on other platforms.

AI-Generated Reviews

How Much of the Problem Is AI?

This is the fastest-moving part of the fake review landscape, and several platforms are now reporting AI-specific figures for the first time.

Yelp filtered nearly 500,000 suspected AI-generated reviews in 2025 out of roughly 22 million total reviews contributed. AI-generated reviews violate Yelp’s content guidelines regardless of intent, because all reviews must be based on genuine firsthand experience.

Tripadvisor removed 214,000 AI-generated reviews in 2024, citing the risk of consumers encountering a “sea of sameness” rather than real firsthand insight. These are the only two major platforms currently publishing discrete AI-review removal figures.

Pangram Labs analyzed nearly 30,000 front-page reviews across 500 best-selling Amazon products and found 909 AI-generated reviews, about 3% of the total. The rate was higher in certain categories, rising to about 5% in baby products, beauty, and wellness. Of those AI-written reviews, 74% gave five stars and 93% carried the verified purchase label (2025). Note: Pangram Labs sells AI-detection software.

The Transparency Company put AI-generated reviews at 3.1% of all detected fakes in a 73-million-review study, growing at roughly 80% month over month since June 2023 (2024). At this point, AI-generated content is a real and growing portion of the fake review ecosystem, but the majority of fake reviews are still written by humans.

One important distinction that’s easy to miss: AI-written and fraudulent aren’t the same thing. A non-native English speaker using ChatGPT to clean up a genuine review of a product they actually bought isn’t committing fraud, even though Yelp’s policy prohibits it. Removal counts for AI reviews reflect a policy choice, not confirmed deception.

Why “Spot the Fake” Advice Rarely Works

Plenty of articles offer tips for identifying fake reviews. Look for generic language, excessive enthusiasm, lack of specific detail. These might have worked five years ago, but they don’t work now.

In a HICSS 2025 study, 151 consumers were asked to classify reviews as genuine or AI-generated across 906 total classifications. Their accuracy rate was 53.2%, which is essentially random. They performed even worse on negative AI-generated fakes.

The UCLA Anderson study backs this up from a different angle. In an incentive-compatible survey (respondents were paid more for correct guesses), people rated products that were actively cheating with fake reviews at 42% suspicion and honest products at 39%. The difference was statistically negligible. Consumers couldn’t tell them apart.

That’s why platform detection has shifted from text analysis to network analysis. A network-based method published in PNAS that targets the product-reviewer graph rather than the review text identified fake review buyers with 86% accuracy and an AUC score of 0.93. Better-written fakes don’t evade this approach because the detection is based on who reviewed what, and when, rather than what the review says.

What Platforms Are Removing

The major platforms now publish annual enforcement figures. Here’s the latest from each.

Google Maps (2025): Blocked or removed more than 292 million policy-violating reviews. Removed or blocked more than 13 million fake Business Profiles. Blocked 79 million inaccurate or unverified edits. Placed posting restrictions on more than 782,000 policy-violating accounts. Also published more than 1 billion helpful reviews.

Amazon (2025): Proactively blocked “hundreds of millions” of suspected fake reviews. Shut down more than 100 websites facilitating fake reviews. More than 40 fake review brokers ceased activity as a result of legal action.

Trustpilot (2025): Removed 7.8 million fake reviews from 361 million total active reviews. All submitted reviews (roughly 200,000 per day) are screened by automated detection before publication.

Trustpilot (2024): Removed 4.5 million fake reviews, amounting to 7.4% of the total submitted, up from 6.1% in 2023. 90% were caught by automated detection.

Yelp (2025): Filtered nearly 500,000 suspected AI-generated reviews. Closed more than 1.3 million accounts for policy violations, up 138% from 2024.

Tripadvisor (2024): Rejected or removed more than 2.7 million fraudulent reviews, up from 2 million in 2023. Of 31.1 million total reviews, 87.8% cleared automated screening, 7.3% were rejected by technology, and 4.9% were flagged for human moderation.

Why Removal Counts Can Be Deceptive

These numbers are impressive in isolation, but they’re limited in what they can tell you. A rising removal count can mean the platform is getting better at catching fakes. It can also mean the problem is getting worse. These reports don’t allow you to distinguish between the two.

The figure that actually measures how well enforcement works is the detection lag. And the data on that isn’t encouraging. The Marketing Science study measured it directly: a fake review campaign runs for a median of six days. Amazon takes an average of more than 100 days to delete the resulting reviews. By that point, the sales boost has already happened.

One other detail worth noting: Amazon’s reporting changed between its 2024 Brand Protection Report, which gave a specific figure of more than 275 million blocked reviews, and the 2025 Trustworthy Shopping Experience Report, which says “hundreds of millions.”

Laws and Enforcement

The United States

The FTC Rule on the Use of Consumer Reviews and Testimonials (16 C.F.R. Part 465) took full effect on October 21, 2024. It prohibits:

• Creating, buying, selling, or disseminating fake reviews
• Paying for reviews conditioned on a particular sentiment, positive or negative
• Undisclosed insider reviews from employees or their immediate relatives
• Company-controlled “independent” review sites
• Certain forms of review suppression
• Buying fake indicators of social media influence

Penalties can reach up to $53,088 per violation for violations.

The FTC’s enforcement trajectory has moved quickly. In December 2025, the agency sent warning letters to 10 companies over potential violations, the first public enforcement step under the rule. The FTC published the actual warning letter template, which is unusual and worth reading.

In May 2026, the DOJ sued Premium Home Service on the FTC’s behalf, alleging the company created thousands of fake local business listings for home repair companies with fabricated five-star reviews written by employees, relatives, and SEO contractors. Minnesota filed a parallel state action the same day. A final order in July 2026 included a $4 million judgment. FTC Commissioner Mark Meador’s concurring statement focused on the harm to honest competitors, calling out how the fraud “defeated the efforts of diligent consumers who went to extra lengths to try to hire reputable, local providers with genuine reviews.”

The United Kingdom and EU

Under the Digital Markets, Competition and Consumers Act 2024, posting or commissioning fake reviews, publishing undisclosed incentivized reviews, concealing negative reviews, and presenting misleading star ratings became automatically unfair banned practices from April 2025. Platforms also have a positive duty to take reasonable and proportionate steps to prevent and remove fake reviews. The CMA can fine up to 10% of global turnover.

A July 2025 sweep of more than 100 review-publishing businesses found over half potentially non-compliant. The CMA issued 54 advisory letters, and 90% of recipients made changes in response.

In March 2026, the CMA opened five formal investigations into Autotrader and Feefo (for suppressing one-star reviews), Dignity Funerals (for staff-written reviews), Just Eat (for inflating star ratings), and Pasta Evangelists (for undisclosed incentivized reviews). Those cases are still open.

The two approaches differ in an important way. The FTC model penalizes per violation after the fact. The UK’s CMA imposes an ongoing positive duty on platforms and can fine based on a percentage of global turnover. The per-violation math can add up fast for a company with thousands of fake reviews, but the turnover-based ceiling hits harder at the top.

What Consumers Think

BrightLocal’s 2026 Local Consumer Review Survey of 1,002 US adults provides a useful snapshot of where consumers stand.

The headline finding is that consumer trust in reviews hasn’t collapsed. 97% still read reviews before choosing a local business, even with widespread awareness of fake content. What’s shifted is the expectation that somebody will do something about it.

93% think someone should be responsible for detecting fake reviews. That accountability is split, with 63% pointing at review platforms, 49% saying businesses themselves, and 25% saying government or legal authorities. Half of respondents think more than one group should share the responsibility, and 5% think all four (platforms, businesses, consumers, and government) should be involved.

The vast majority of respondents want consequences. 97% think businesses caught using fake reviews should face some form of punishment. 37% support financial penalties. 46% want businesses removed from Google search results. 16% favor criminal charges or jail time.

Younger adults (18 to 29) are the most lenient. The strictest group is consumers aged 45 to 60. And consumers who’ve regretted a purchase of more than $5,000 after relying on reviews are the most likely to support harsh penalties.

Frequently Asked Questions

How many online reviews are fake?

Published estimates from government research and academic studies range from about 4% to 16% of reviews platform-wide. The 11% to 15% range from the UK Department for Business and Trade and the roughly 14% finding from a 73-million-review Google analysis are the most methodologically defensible figures available. Higher estimates exist but rely on proprietary detection methods.

Are fake reviews illegal in the United States?

Yes. The FTC Rule on the Use of Consumer Reviews and Testimonials, effective October 21, 2024, prohibits creating, buying, selling, or disseminating fake reviews. It also prohibits paying for reviews conditioned on a particular sentiment and undisclosed insider reviews. The FTC has begun enforcement actions under the rule, including warning letters and lawsuits.

What is the penalty for fake reviews?

Under the FTC rule, civil penalties can reach $53,088 per knowing violation. In the UK, the CMA can fine up to 10% of global turnover under the DMCC Act. The FTC’s first enforcement action under the rule, the Premium Home Service case, resulted in a $4 million judgment.

Can you tell if a review is fake?

Consumers generally can’t. A HICSS 2025 study found people identified AI-generated fake reviews correctly only 53.2% of the time. And in a UCLA Anderson study using incentive-compatible methods, respondents assigned nearly identical suspicion scores to products that were cheating (42%) and products that were honest (39%).

Do verified purchase badges mean a review is real?

A verified purchase badge confirms that a transaction occurred. It does not confirm that the review is genuine. 93% of AI-written reviews on Amazon carried the verified purchase badge, and the most common fake review scheme involves a recruited reviewer making a real purchase and getting reimbursed through PayPal after posting their review.

What should a business do about fake negative reviews?

If fake negative reviews appear on your profile, flag each one through the platform’s reporting tools. If the reviews are accompanied by a payment demand, do not engage or pay. Document everything, including screenshots and messages, and file through Google’s dedicated extortion form if the reviews are on Google. Post a brief, professional note on your profile explaining the situation while waiting for the reports to be processed.