AI Transparency and User Explanation Policy

How to use this document

• What it is for. This policy defines what customers, users, candidates and other affected people are told about the use of AI, and the circuit for handling their explanation requests: which channel they come through, who answers, within what time, with which response model, and where each explanation delivered is recorded.
• What it does not cover. The technical way to obtain a model's factors is in the "Model Explainability (XAI) Procedure"; this document takes that information and turns it into an answer the person can understand. Human review, contesting and opting out of automated decisions are handled under the "Automated Decision Review Procedure". The catalog of rights communicated to people is in the "AI User Rights Notice".
• Who should own it. Usually the AI lead, with Legal, the Data protection officer (or the country's equivalent role) and the owner of each system. Enter the function in [RESPONSIBLE FUNCTION] and the person in [OWNER].
• Decisions to make first. Decide which transparency level applies to each system (section 5 and aid E), which internal response time you will give to explanation requests (section 10 and variants), who answers for each system, and in which languages and accessible formats notices will be offered.
• Fields to complete. Replace the bracketed text, complete the table in aid E with your systems, adapt the models in aid G and set up the register in aid H. Keep [OPTIONAL] clauses only if they apply.
• Country variants. Keep the variants for the countries where you serve people and delete the rest. Each variant states what changes: minimum content, related rights and deadlines set by the law itself.
• Review. Ask Legal and the Data protection officer to validate the assigned levels, deadlines and texts, especially for decisions with legal or similarly significant effects.
• Approval and publication. Have the AI committee approve it, record the approval in "Version history", communicate it to product, design, marketing and customer service, and train customer service staff before applying it.

AI Transparency and User Explanation Policy

Field	Value
Organization	[LEGAL ENTITY NAME]
Owning function	[RESPONSIBLE FUNCTION]
Policy owner	[OWNER]
Approved by	[APPROVED BY]
Effective date	[DATE]
Review frequency	[REVIEW FREQUENCY]
Main channel for explanation requests	[PRIVACY EMAIL]
Default internal response time	[COMPLETE: business days]

1. Purpose

1.1 [ORGANIZATION NAME] wants people to trust how it uses artificial intelligence. To do so, they must know when they are dealing with an AI system, when content has been generated or altered with AI and when AI influences a decision that affects them, and be able to understand that decision.

1.2 This policy sets out what we disclose, when, how and in how much detail, in proportion to the risk of each use, and how the circuit works through which a person requests and receives an explanation.

2. Scope

2.1 This policy applies to all AI systems that [ORGANIZATION NAME] develops, integrates or uses and that:

a) interact directly with external people, such as conversational assistants, voice assistants or agents;
b) generate or manipulate image, audio, video or text content that is published or delivered to third parties;
c) influence outcomes or decisions about external people, such as customers, users, candidates, patients or students;
d) detect emotions or categorize people based on biometric data, where the law allows it.

2.2 [OPTIONAL] This policy also applies to AI systems that affect workers, with the adaptations in the "AI and Algorithm Notice to Workers, Candidates and Representatives".

2.3 It applies to all staff who design, develop, communicate about or handle inquiries on these systems, and to vendors that operate them on behalf of the organization.