Hoozright
AI Policy
Version 1.1 · Last reviewed 25 April 2026
INTRO SECTION — OUR APPROACH TO AI
Hoozright uses artificial intelligence in specific, defined ways to operate the Platform. This policy explains where AI is used, what it does, what it does not do, and what safeguards are in place to protect users.
Hoozright’s core value proposition is that judgement matters — that being right, early, and consistent should be recognised and rewarded. We hold ourselves to the same standard when it comes to AI: we use it where it genuinely improves the Platform, we are transparent about where and how, and we maintain human oversight over decisions that affect users.
SECTION 1 — WHERE WE USE AI
The table below sets out every use of AI on the Platform as of the date of this policy. We will update this policy before introducing any new AI use.
| AI use | Description |
|---|---|
| Content moderation | Google Gemini analyses comments, debate titles, and submitted evidence (text and links) to detect content that may breach our Community Standards — including toxicity, hate speech, spam, and illegal content. Moderation decisions are flagged for human review before any enforcement action is taken. |
| Debate enhancement | Google Gemini may be used to suggest improvements to debate framing, identify ambiguities in debate titles, or surface related debates. This is assistive only — users are never required to accept AI suggestions. |
| Platform content generation | We use AI to produce certain editorial and contextual content that supports the Platform experience. This includes example debates, category descriptions, onboarding content, and other supporting material intended to illustrate platform features or improve discoverability. This content is reviewed before publication and is clearly distinct from content submitted by users. |
| Reputation Score calculation | The Reputation Score is calculated by our proprietary algorithm, which takes into account vote accuracy, vote timing (early, mid, or late), and debate outcomes. The algorithm is deterministic and rule-based; it does not use a machine learning model to assess individual users. |
| Opinion Intelligence (future) | When activated, this B2B data product will use aggregation and anonymisation techniques to produce structured opinion datasets from Platform activity. No individually identifiable data will be included. See our Privacy Notice for further detail. |
SECTION 2 — WHAT AI DOES NOT DO ON HOOZRIGHT
We believe it is as important to be clear about the limits of AI use as it is to disclose where AI is used. The following decisions are not made by AI on Hoozright:
- Account suspensions and bans — these are reviewed and confirmed by a human
- Final content removal decisions where a user has submitted an appeal
- Determinations about whether a debate outcome has been correctly resolved
- Judge status grants or revocations based on suspected manipulation
- Any decision that results in financial consequences to a user (for example, subscription cancellation for policy breach)
SECTION 3 — THE REPUTATION SCORE
The Reputation Score is central to Hoozright's product. Because it determines your standing on the Platform — including access to Judge status — we want to be transparent about how it works.
3.1 Inputs
Your Reputation Score is calculated from:
- Whether your votes align with debate outcomes (accuracy)
- When you cast your votes relative to other users (timing — early votes that prove correct are weighted more heavily than late votes)
- The volume of debates you have participated in
3.2 What the score does not use
Your Reputation Score is not influenced by:
- Who you are — your identity, demographics, or any personal characteristics
- Your follower count or social connections on or off the Platform
- Whether Hoozright agrees with your opinions
- Any inferred political, social, or ideological alignment
3.3 Algorithm changes
We may update the Reputation Score algorithm from time to time to improve accuracy and fairness. Where a change materially affects how scores are calculated, we will notify users via an in-Platform notification before the change takes effect and explain what is changing and why.
SECTION 4 — CONTENT MODERATION BY AI
4.1 How it works
When you submit a comment, debate title, or piece of evidence, it may be passed to Google Gemini for automated review. Gemini assesses the content against our Community Standards and returns a recommendation. This processing happens server-side and does not involve cookies or client-side tracking.
Content flagged by Gemini is not automatically removed. A human moderator reviews flagged content before any action is taken, except in cases of content that is unambiguously illegal (for example, child sexual abuse material), which is removed immediately and reported to the relevant authorities.
4.2 Accuracy and errors
AI content moderation is not perfect. Gemini may flag content that does not breach our standards (false positives) or miss content that does (false negatives). We are committed to continuous improvement of our moderation processes.
If your content is removed or your account is actioned and you believe this is an error, you have the right to appeal. See Section 6 below.
4.3 Data passed to Google Gemini
Content moderation involves passing user-submitted text and links to Google's Gemini API. This processing is governed by Google's data processing terms, which we have agreed to as part of our use of the Gemini API. We do not pass account identifiers or personal data alongside content for moderation purposes. For full details of how we share data with third parties, see our Privacy Notice.
SECTION 5 — EU AI ACT COMPLIANCE
The EU AI Act establishes a risk-based framework for AI systems used in the European Union. We have assessed our AI uses against the Act's requirements.
| AI use | Risk classification and basis |
|---|---|
| Content moderation (Gemini) | Limited risk. We use Gemini as an assistive tool with human review before enforcement. We do not rely on it as a sole decision-maker. We maintain logs of moderation decisions for audit purposes. |
| Platform content generation | Minimal risk. AI-generated content is reviewed before publication. |
| Reputation Score algorithm | Minimal risk. The algorithm is rule-based and deterministic, not a general-purpose AI model. Its inputs and logic are documented internally. |
| Debate enhancement suggestions | Minimal risk. Suggestions are optional and user-controlled. No consequential decision is made on the basis of these outputs. |
| Opinion Intelligence (future) | We will conduct a full AI Act assessment before activating this product and will update this policy accordingly. |
We are not currently aware of any use of AI on the Platform that falls into the EU AI Act's high-risk or prohibited categories. If our assessment changes, we will update this policy and take appropriate steps to comply.
SECTION 6 — YOUR RIGHTS AND HOW TO APPEAL
If you believe an AI-assisted decision on the Platform has affected you unfairly — for example, content removed in error, a Reputation Score that appears incorrect, or a moderation action you disagree with — you have the right to request human review.
To submit an appeal or query, contact us at: support@hoozright.com
Please include your username, the relevant debate or content, and a description of the issue. We aim to respond to all appeals within 5 business days.
We will not use the fact that you have submitted an appeal as a factor in any future moderation or scoring decision.
SECTION 7 — HUMAN OVERSIGHT COMMITMENT
We maintain internal logs of AI-assisted moderation decisions, including the recommendation made by Gemini and the outcome of any human review. These logs are reviewed periodically to assess accuracy and bias.
We will conduct a review of this policy and our AI systems at least annually, and sooner if we introduce new AI features or if regulatory requirements change.
SECTION 8 — THIRD-PARTY AI SERVICES
Hoozright uses Google Gemini as its AI provider. We do not build or train our own large language models. Our use of Gemini is governed by our agreement with Google, including their data processing and acceptable use terms.
We selected Google Gemini on the basis of its performance on content moderation tasks, Google's data processing commitments, and its compliance with applicable data protection law. We continue to monitor the market and may change AI providers in the future. Where a change in provider would materially affect how user data is processed, we will update our Privacy Notice and notify users.
SECTION 9 — CHANGES TO THIS POLICY
We will update this policy before introducing any new AI use on the Platform. Where changes are material, we will notify registered users by email or via an in-Platform notification. The date at the top of this page reflects the most recent revision.
SECTION 10 — CONTACT
Killadreenan, Newtownmountkennedy, Co. Wicklow, IE
This document requires independent legal review before publishing.