Detection is not enough to stop fake reviews because a fake has to be written and submitted before any filter can act on it, and AI now makes each fake cheap to produce and hard to tell from a real one. Every major platform runs detection, and every major platform still removes fakes by the million. The fix that holds up is structural: accept a review only when it is tied to a real, completed payment.
The stakes are high. 98% of consumers at least occasionally read online reviews when researching local businesses (BrightLocal, 2023).
The numbers
The scale of fake reviews is staggering, and it's growing:
- Trustpilot removed 4.5 million fake reviews in 2024: 7.4% of all reviews submitted to the platform that year, up from 6.1% in 2023. (Trustpilot Trust Report 2025)
- Amazon blocked over 200 million suspected fake reviews in 2022 alone. (Amazon)
- Google blocked or removed over 292 million policy-violating reviews on Maps in 2025. (Google)
- A 2021 CHEQ analysis published by the World Economic Forum put the direct influence of fake reviews on global online spending at $152 billion a year. (World Economic Forum)
- 80% of consumers believe they've read a fake review in the last year. (BrightLocal, 2020)
For the full statistical breakdown, see our Fake Review Statistics 2026 page.
Why AI makes the problem worse, not better
Before 2023, fake reviews were often easy to spot: broken English, vague praise, repetitive phrasing. Detection systems looked for these patterns.
Generative AI erased those signals. A single prompt to ChatGPT or Claude can generate hundreds of unique, grammatically perfect, emotionally nuanced reviews: each with different vocabulary, sentence structure, and level of detail. In two experiments published in Marketing Letters in April 2024, researchers found that "humans cannot recognize AI-written reviews," and AI detectors were fooled too.
Platforms can still detect fake-review operations: coordinated campaigns, IP clusters, review velocity anomalies. But the individual review? If it's well-prompted and one-off, no algorithm can reliably tell it apart from a genuine one.
The detection arms race is now asymmetric. Generating a convincing fake review costs almost nothing and takes seconds. Detecting one, if it's well-crafted, costs far more in compute, human review time, and false-positive risk. The economics favor the fakers, and AI is widening the gap.
The regulatory response
Governments are beginning to act, but regulation is inherently slower than technology. Here's where fake review law currently stands across the major markets.
United States: FTC's 2024 Trade Regulation Rule
On 21 October 2024, the Federal Trade Commission's Trade Regulation Rule on the Use of Consumer Reviews and Testimonials (16 CFR Part 465) took effect (FTC Q&A). According to the FTC's announcement, the rule:
- Prohibits reviews that misrepresent that the reviewer had genuine experience with a product or service
- Bans buying or selling reviews, including "incentivized" reviews where the incentive is conditioned on sentiment
- Prohibits undisclosed insider reviews: officers, managers, employees or agents reviewing without disclosing their connection
- Bans review suppression, threatening or intimidating reviewers to remove negative reviews
- Lets courts impose civil penalties for knowing violations
The rule is strong on paper. But enforcement requires detection: the FTC must find the fake reviews to penalize them. And detection, as we've established, is a losing battle.
United Kingdom: Digital Markets, Competition and Consumers Act
Since April 2025, the UK's DMCC Act treats these review practices as banned, meaning automatically unfair and illegal (CMA):
- Obtaining and posting fake reviews
- Paid-for reviews that are not clearly marked as incentivized
- Hiding negative reviews, or star ratings that present an inaccurate picture
The Competition and Markets Authority (CMA) can fine businesses up to 10% of global turnover.
European Union: Digital Services Act
The DSA requires "very large online platforms" to assess and mitigate systemic risks, including the spread of illegal content and disinformation. Fake reviews, when they distort consumer decisions at scale, fall under this mandate. Platforms face fines of up to 6% of global annual turnover (European Commission).
The SiteJabber precedent
In November 2024, the FTC charged SiteJabber, a consumer review platform, with inflating ratings using reviews from people who had not yet received the products they reviewed. According to the FTC, SiteJabber collected ratings at the point of sale, before customers received what they bought, and counted them as if they reflected real experience with the product.
The FTC approved a final order in January 2025. It bars SiteJabber from misrepresenting that its ratings and reviews came from customers who had actually received the products. The company neither admitted nor denied the allegations. The FTC framed the case alongside its new rule: "cases like this one show that we'll act to stop all forms of deception in the review ecosystem."
The SiteJabber case illustrates the structural point: when verification depends on what the merchant says, the system can be gamed, even unintentionally, at scale.
Why detection is structurally insufficient
Every detection-based approach shares the same flaw: a fake review must be written before it can be caught. By the time a platform identifies and removes a fake review:
- It may have been live for days, weeks, or months
- It may have influenced hundreds or thousands of purchasing decisions
- The business that posted it may have already moved on to the next batch
- The platform's reputation, and the reputation of every honest business on it, has already taken the hit
Detection is a treadmill. The faster you run, the faster the ground moves beneath you. AI accelerates the ground speed; regulation adds a slight incline. Neither changes the fundamental dynamic.
Understanding the mechanics does help you catch the amateur attempts. See how fake reviews work for the playbook, and you can run any suspicious review through our free fake review checker to score it against the common signals.
The structural alternative: processor-attested verification
There is one way to make fake reviews structurally expensive: require every review to be tied to a real, completed payment recorded by the payment processor.
This is processor-attested verification, and it's the only review verification method where the attesting party is independent of both the merchant and the reviewer:
| Approach | Who attests | Can the merchant fake it? |
|---|---|---|
| Reactive detection | No one (algorithm hunts fakes) | Yes, until caught |
| Merchant-supplied (Level 3) | The merchant | Yes, requires only a fake order record |
| Processor-attested (Level 4) | The payment processor | Only by making a real Stripe charge, which costs real money and risks account closure |
For the full verification spectrum (Levels 0 through 4, with what each platform actually does), see What Does "Verified Buyer" Actually Mean?.
How it works
- A customer completes a purchase through Stripe.
- Signed Reviews picks up the new charge (read access) and creates a unique, expiring review invitation tied to that specific transaction.
- The invitation is sent to the email address on the Stripe payment, the customer's verified payment email.
- The customer clicks, writes their review, and submits it. The review is cryptographically signed at submission.
- If the charge is fully refunded or disputed, a background check that re-reads charges every 6 hours hides the review automatically.
No purchase → no review. This isn't detection. It's prevention. Someone who never bought has no path to posting a review. Faking one means paying for a real charge first.
And critically, the charge record behind each review comes from Stripe, a regulated financial institution and an independent third party to every transaction, not from the merchant. Even if a merchant wanted to game the system, they'd need to create real Stripe charges (paying Stripe's fees each time) and risk having their Stripe account closed for suspicious activity.
What this means for your business
If you're an honest business, the fake-review problem costs you in three ways:
- Unfair competition. A competitor with 500 AI-generated 5-star reviews outranks your 50 genuine 4.5-star reviews: on Google, on review platforms, and in consumer trust.
- Review extortion. Customers increasingly use the threat of negative reviews to demand refunds or discounts, knowing platforms rarely remove reviews, and that your genuine rating is fragile.
- Trust erosion. When consumers can't tell real reviews from fake ones, they trust all reviews less. Your genuine reviews lose value because the category itself is tainted.
Processor-attested verification solves all three: competitors can't post reviews against your business without buying from you, extortion threats lose force because your rating is built on verified purchases, and your review portfolio stands out as tied to real payments in a sea of unverifiable noise.
The bottom line
Fake reviews are a structural problem, not a moderation problem. You don't fix a structural problem with better detection. You fix it with a structure that prevents the problem from existing.
Processor-attested verification is that structure. It makes fake reviews costly by design, not just detectable after the fact. And for businesses on Stripe, it's available today: one click, no code, free to start.
Further reading: Stripe Verified Reviews explains how processor-attested verification works in practice. Fake Review Statistics 2026 has the full data. What Does "Verified Buyer" Actually Mean? breaks down what the badge means on every major platform. For a practical 7-point checklist for evaluating any review, see our fake review checker guide. The company behind this approach is Paid Rightly LLC, a New Mexico limited liability company.
Every review tied to a real Stripe payment
Signed Reviews attaches a cryptographic signature to each review, linked to a completed transaction. Free to start: unlimited self-service reviews, 10 automated invitations, no credit card.
Start free → How verification works