Purpose & objectives

The purpose of this policy is to ensure that users interacting with AI-driven systems operated or deployed by Zen AI Governance UK Ltd are:

• Made aware when they are engaging with or being affected by AI.
• Given appropriate information to understand AI capabilities, limitations, and oversight measures.
• Enabled to exercise their rights to human review, rectification, or appeal of AI-driven outcomes.

Transparency principles

1. Disclosure: Inform users whenever AI contributes to a decision, recommendation, or response.
2. Comprehensibility: Use plain language and contextual explanations suitable for the audience.
3. Accessibility: Provide disclosures in all accessible formats used by the platform (web, voice, chat, document).
4. Timeliness: Offer transparency before or at the point of interaction, not retroactively.
5. Traceability: Link all disclosures to the underlying system record, dataset, and version.

Disclosure requirements

• All user-facing AI systems must include a **Transparency Notice** prominently displayed at the first point of interaction.
• Notices must include:
  • System name and version.
  • Purpose and role of AI in the process.
  • Human oversight and escalation route.
  • Data sources and model update frequency.
  • User rights: review, complaint, or opt-out (if applicable).
• Disclosure language must align with accessibility and equality standards (WCAG 2.2 AA).

Disclosure formats & templates

Zen AI Governance provides standard templates to ensure uniform transparency across all systems:

⚠️ This response is generated by an AI system (ZenBot v2.4) trained on verified regulatory sources.
All outputs are reviewed by a qualified compliance specialist before publication.
If you wish to speak to a human advisor, please click “Request Human Review”.
  

Explainability & model communication

• AI systems must include a documented **Explanation Type** (e.g., feature-based, example-based, surrogate model).
• For each high-risk AI system, maintain a **User-Facing Explanation Summary (UFES)** detailing:
  • Inputs influencing outputs (feature importance).
  • System confidence score or uncertainty indicator.
  • Limitations, assumptions, and risk mitigations.
• Explanations must be interpretable by non-technical audiences (plain-language summaries).

User rights & contestability

• Notification: Users must be informed of AI use prior to engagement.
• Explanation: Users have the right to a clear, concise explanation of how an outcome was generated.
• Human review: On request, a human decision-maker must reassess an AI-based decision.
• Appeal & rectification: Provide a process to challenge outcomes through an escalation channel.
• Data rights: Users may request access, correction, or deletion of their data under the UK GDPR / DPA 2018.

Disclosure channels & timing

• Web & App Interfaces: Persistent info button + popup disclosure before data input.
• Voice Systems: Spoken notice: “This call is assisted by AI.”
• Email / Document Automation: Footer statement: “Generated with AI review.”
• In-person Kiosk / Chatbot: Visual indicator and opt-out route provided.

Regulatory alignment

Example transparency notices

• Web Chatbot: “Our assistant uses AI to generate replies. All messages are reviewed for accuracy and compliance.”
• Document Generator: “This report was assisted by AI. Verify factual accuracy before final submission.”
• Decision Support Tool: “AI is used to prioritise applications. A human reviewer validates all final decisions.”

Implementation checklist

• Transparency notices created and reviewed for all AI interfaces.
• Explanation Summaries (UFES) approved by AI Governance Board.
• Disclosure templates localised for accessibility and languages.
• Contestability workflow active in CRM/Helpdesk channels.
• Annual transparency review conducted by Compliance Lead.

© Zen AI Governance UK Ltd • Regulatory Knowledge • v1 15 Nov 2025 • This page is general guidance, not legal advice.