Right to explanation of AI and algorithmic decisions

The right to explanation follows indirectly from GDPR Art. 13, 14, 15 (information duty + access right) combined with Recital 71. For automated decision-making the controller must provide meaningful information about the underlying logic + significance + envisaged consequences of the processing. What is "meaningful"? Not source code (no human would understand that), but: which factors were weighed, what weight each factor had, what data sources are used, what outcome ranges are possible, and how a human would make the same decision. E.g. for a refused mortgage: "we weigh income (40%), debts (25%), credit registration (20%), employment history (15%); your application scored below our threshold due to X and Y". Three info moments: (1) Up front — privacy statement must mention that automated decisions occur and on what. (2) On request (Art. 15) — you can always ask whether automated decisions were made about you and what logic was used. (3) On the decision itself — obtain explanation of the concrete decision. What if they cite "trade secret"? Not absolute. EDPB guidance: business interest is balanced against your right; complete silence is not allowed. AI Act from 2026: high-risk AI systems get an additional explanation duty (Article 86) — users receive a "clear and meaningful explanation" of how the system contributed to the decision. Practical: combine Art. 15 + Art. 22 in one request: "I request access to all data processed about me, and meaningful explanation of the underlying logic for the decision of date X on subject Y". 1-month response time.