For years, online reviews shaped purchasing decisions through a predictable path: the consumer searched on Google, found the review platform's page, read the opinions, and decided. The brand could monitor that path and intervene when necessary.
That path is changing. ChatGPT, Google AI Mode, Perplexity and Gemini no longer show a list of links. They give a direct answer, with cited sources. And among the sources they cite, review platforms hold a central position.
The data: 800,000 AI answers analysed
A study published in May 2026 by Seer Interactive analysed 804,491 responses generated by four AI platforms (ChatGPT, Google AI Mode, Perplexity, Google Gemini), covering 1,926 brands across eight different verticals. The goal was to measure how much review platforms influence brand visibility in AI answers.
The result: review platforms are the second most cited source in AI answers, after generic websites. The most relevant finding is how this influence shifts along the purchase journey. In the exploration phase, reviews account for 1.5% of citations. In the purchase decision phase they rise to 24.3%: a sixteen-fold increase. When the consumer is ready to buy, the AI uses reviews as final validation.
Absence, manipulation, amplification
The study documents a new phenomenon: when a brand has no profile on a review platform, AI engines explicitly flag this as an indicator of lower trustworthiness. Absence is no longer neutral. It is an active penalty.
The mechanism works in reverse too. If the published score is the result of a manipulated collection, that figure is cited and redistributed by every AI engine, across every channel, simultaneously. The error at the source propagates everywhere. The same happens with a ratings drop or a regulatory ruling: the correction does not stay confined to the platform of origin.
In short: any reputational signal, positive or negative, authentic or constructed, is now amplified by the AI ecosystem. The question is whether that signal has integrity.
The current blind spot and the correction ahead
Today, AI engines cite review platform scores with no critical capacity. They take the published number and repeat it. They do not distinguish a score built through organic collection from one inflated with selective invitations or incentives. This blind spot is real, and it is why score manipulation still works.
But the context is shifting on two fronts.
The first is Google. Google owns Google Reviews: every review, every user profile, every timestamp, every behavioural pattern. Google AI Mode and Gemini have access to all of this data. Google already removes reviews through automated filters today. The step from removing fake reviews to flagging in AI answers that a profile shows anomalies is technically short. For Google it is not a question of capability, it is a question of priority.
The second is regulatory pressure. In the United Kingdom, the Digital Markets, Competition and Consumers Act 2024 came into force on 6 April 2025, making fake reviews, undisclosed incentivised reviews, the suppression of negative reviews and misleading aggregate ratings banned practices, with fines of up to 10% of global turnover. The Act also places a positive obligation on businesses that publish a third-party aggregate score, a Trustpilot rating being the explicit example, to maintain a public reviews policy. The Competition and Markets Authority moved from guidance to active enforcement in July 2025. At European level, the Digital Services Act and rulings such as the Italian competition authority's 4 million euro sanction against Trustpilot are pushing platforms towards greater transparency on collection processes. As integrity data becomes structured and accessible, every AI engine will be able to integrate it into its answers.
When this happens, and it is a question of when, not if, brands with manipulated collection will face a simultaneous correction across all AI channels. No regulatory ruling will be needed. The information ecosystem itself will downgrade them. And unlike a sanction, which is a discrete and appealable event, algorithmic correction is permanent and non-negotiable.
The implications for businesses
Reputational risk is no longer a problem confined to a single platform. It is an ecosystem problem. A brand with non-integral review collection is polluting its informational presence across all AI channels at once, often without knowing it.
For CROs and compliance teams, this changes the nature of the risk. It is no longer about monitoring one platform. It is about verifying that the data at the source has integrity, because that data is amplified and distributed automatically across channels the company does not control.
Measuring the integrity of review collection, today, is not an academic exercise. It is an operational necessity. Brands that invest in collection integrity are building an asset. Those that manipulate are accumulating reputational debt that will, sooner or later, be called in.
Read also: Google makes AI visibility measurable. For those who manage reviews, this changes everything.
© Fametrue 2026. All rights reserved.
Back to the Observatory