Right to explanation response template
When a user asks why an automated system rejected their loan, flagged their account, or scored their application, you may have a legal duty to respond with meaningful information about the decision. Vague or evasive replies create compliance and trust risk. This tool drafts a structured, plain-language response that explains the outcome, the key factors, and the user’s right to human review and appeal.
The legal background: what triggers the right to explanation
GDPR Article 22 gives individuals the right not to be subject to a decision based solely on automated processing, including profiling, where that decision produces legal or similarly significant effects. Where such decisions are permitted (under explicit consent, contractual necessity, or EU/member state law), the controller must implement safeguards — including the right to obtain human intervention, express their point of view, and contest the decision.
“Meaningful information about the logic involved” is the key phrase. Courts and regulators have interpreted this to require something between a full technical explanation (which is not required) and a useless non-answer (which fails the standard). In practice this means: what type of model or rule made the decision, which factors were most significant, how those factors were weighted in this individual’s case, and what the person can do if they disagree.
The right applies to decisions in scope of Article 22 — particularly: credit scoring, insurance underwriting, automated job application screening, fraud detection leading to account restrictions, and content moderation resulting in significant consequences. If your system makes any of these decisions, you need a response process.
What “meaningful information” means in practice
Regulators have been clearer over time about what falls short:
- Too vague: “Your application was assessed by our automated system based on various factors.” This tells the person nothing actionable.
- Too technical: Providing feature weights, model architecture, or training data provenance — this is not required and often exposes trade secrets unnecessarily.
- Adequate: “The primary factors in this decision were your payment history on existing credit accounts (weighted most heavily), the number of credit applications in the past 12 months, and your current debt-to-income ratio. Your payment history over the past 24 months showed two missed payments, which had the most significant effect on this outcome.”
The template generates responses structured around the “adequate” standard — factual, specific enough to be actionable, but not disclosing proprietary model internals.
How it works
You enter the decision type, a short description of the AI system, the outcome the user received, and the main factors that influenced it. The tool assembles these into a complete response letter: an opening that acknowledges the request, a plain-language summary of how the decision was made, the key factors and how they were weighed, a statement of significance and consequences, and a clear human-review and appeal path. You then add your contact details and statutory deadline and send after review. Everything is generated locally.
Tips and notes
- Explain factors, not the full algorithm. “Meaningful information” means the logic and significance, not source code or trade secrets.
- Avoid jargon. The reader is usually not technical; plain language is part of the legal standard.
- Always offer human review. Article 22 decisions require a route to contest and obtain human intervention.
- Mind the deadline. Requests usually carry a statutory response window; track it from the date received.
- Have DPO or counsel review. Automated-decision rules vary by jurisdiction and case; the template provides structure but the specific content must be reviewed before sending.