Installing the extension and browsing product pages sends us no data at all. Reading the review text, scoring it, and displaying the verdict all happen and end inside your browser.
We do not collect account details, browsing history, IP addresses, product names, review text, or information about who wrote a review.
Next to each verdict there is a question — "Is this verdict right?" — with Correct / Wrong buttons. Clicking one sends the items below to our server. If you do not click, nothing is sent.
| Field | Contents | Why we need it |
|---|---|---|
user_label | Your judgement (genuine / fake) | The ground-truth label |
engine_verdictengine_scoreengine_version | What our engine decided, its score, and its version | To see where the engine is wrong |
features | Numeric features — character count, how many numbers were mentioned, how many component words, sensory expressions, complaints, colloquial expressions, the formal-register ratio, the highest similarity to another review on the page, the detected language, and so on | Training data |
platform | Which platform (e.g. coupang, amazon) | Accuracy per platform |
ui_language | Your browser's language setting | Accuracy per language |
session_hash | A random string that lives only while the browser tab is open. It disappears when you close the tab and a new one is generated next time. | Detecting bulk injection of fake labels |
The features we receive are numbers like "length 47, two numeric mentions, one complaint expression". The original sentence cannot be reconstructed from them. We also do not receive who wrote the review or which product it was for.
session_hash is randomly generated and is not linked to your account, device, or IP. It disappears when you close the tab. Its only purpose is to catch patterns such as "hundreds of submissions from one tab in a short time". Without that defence, anyone could poison the dataset and break the service for everyone.
When a brand or manufacturer uses our separate audit report API, submitted review text is scored immediately and only the score is stored — the text itself is not stored (it is identified by a SHA-256 hash only). The email address used to obtain an API key is used solely to provide the service and answer enquiries.
We do not disclose your information to third parties. We do rely on the infrastructure providers below, which only store and transmit data on our behalf.
| Processor | Role | Location |
|---|---|---|
| Railway Corp. | Application server and database hosting | United States |
| Cloudflare, Inc. | Website delivery | Global CDN |
This means the information described in section 2 may be stored and processed outside your country, including in the United States. The categories, purposes, and retention periods are exactly those listed in section 2.
You may exercise the following rights at any time by contacting us at the address in section 10.
One thing we should say plainly. Feedback data contains nothing that identifies you. So if you ask us to "find and delete only what I sent", we cannot tell which records are yours. That is a deliberate consequence of designing the system to hold no personal data. If you want data removed, tell us the approximate time and platform and we will review and act on that range.
If you are in the European Economic Area or the United Kingdom, you have the right to lodge a complaint with your local supervisory authority. Users in the Republic of Korea may contact the Korea Internet & Security Agency Privacy Center (privacy.kisa.or.kr, 118) or the Personal Information Dispute Mediation Committee (kopico.go.kr, 1833-6972).
This service is not directed at children under 14 (under 16 in parts of the EEA), and we do not collect age information.
Privacy enquiries and rights requests: projectteamforyou@gmail.com
If this policy changes we will publish the new version on this page and update the dates above. For material changes — such as new fields being transmitted — we will notify you separately when the extension updates.