EU AI Act Article 86: The Right to Explanation of Individual Decisions

What Article 86 Covers

Article 86 of Regulation (EU) 2024/1689 is one of the few provisions of the AI Act that hands a right directly to individuals. Most of the regulation disciplines providers and deployers; Article 86 gives the person on the receiving end of an AI-assisted decision a legal claim to an explanation. It sits in Chapter IX Section 4 on remedies, alongside the Article 85 right to lodge a complaint with a market surveillance authority.

The Conditions of the Right

Article 86(1) defines the trigger with several cumulative conditions. The right applies to any affected person subject to a decision which is taken by the deployer on the basis of the output from a high-risk AI system listed in Annex III, with the exception of systems listed under point 2 thereof, critical infrastructure, and which produces legal effects or similarly significantly affects that person in a way that they consider to have an adverse impact on their health, safety or fundamental rights.

Unpacked, that means: there must be a decision, not merely an AI output; the decision must be taken by a deployer on the basis of high-risk Annex III system output; and it must have legal or similarly significant effects the person considers adverse. Classic examples include credit refusals, insurance pricing decisions, rejection for a public benefit, an adverse recruitment outcome or an educational admission decision, where a high-risk system contributed to the result.

What the Person Can Demand

The affected person has the right to obtain from the deployer clear and meaningful explanations of the role of the AI system in the decision-making procedure and the main elements of the decision taken. Two components, both practical. The role explanation addresses how the system entered the process: what it assessed, what weight its output carried, and what humans did with it. The main elements explanation addresses the decision itself: the principal reasons the outcome went the way it did.

Clear and meaningful is the quality bar. A copy of model documentation is not an explanation, and neither is a generic statement that an algorithm was involved. The deployer must translate the decision logic into terms the affected person can actually use, including to contest the decision through the available remedies.

Limits and Relationship to Other Law

Article 86(2) confines the right: it applies only to the extent that the right is not otherwise provided for under Union law, positioning Article 86 as a gap-filler. Article 86(3) adds that the right applies without prejudice to exceptions or restrictions under Union or national law, which preserves, for example, confidentiality regimes in defined contexts.

The closest neighbour is the GDPR. Article 22 GDPR restricts solely automated decisions with legal or similarly significant effects, and Articles 13 to 15 GDPR give data subjects information about the existence of automated decision-making and meaningful information about the logic involved. Article 86 of the AI Act reaches further in one dimension: it covers decisions taken by the deployer on the basis of AI output, including where a human made the final call, and it targets the specific decision rather than the system in general. Where GDPR rights already supply the explanation, Article 86 does not duplicate them.

How Deployers Should Prepare in Practice

1. Map decision points: identify every process where output from a high-risk Annex III system feeds decisions about individuals with legal or similarly significant effects.
2. Capture decision context at the moment of decision: system version, output values, the human reviewer's action and the principal reasons recorded. The Article 12 logs and the interpretability information that providers must supply under Article 13 are the raw materials.
3. Design the explanation product: a template per decision type stating the role of the AI system in the procedure and the main elements of the decision, written in plain language and reviewed for accuracy against the captured context.
4. Build the request channel: intake, identity verification, response workflow, deadlines aligned with any sectoral rules, and integration with the existing GDPR data subject request process to avoid parallel machinery.
5. Close the loop with Article 27: deployers required to perform a fundamental rights impact assessment must already describe complaint and governance arrangements, and the explanation workflow belongs in that description.
6. Push requirements upstream: procurement should demand from providers the technical capability, flagged in Article 13(3)(b), to provide information relevant to explaining output, since a deployer cannot explain what its supplier made opaque.

A Concrete Example

A bank uses a high-risk AI system under Annex III point 5(b) to evaluate creditworthiness. An applicant is refused a loan after a credit officer reviews the system's score and endorses the recommendation. The applicant, considering the refusal adverse to their fundamental rights, invokes Article 86. The bank's response explains the role of the system, an automated creditworthiness score that the officer used as a primary input and confirmed after file review, and the main elements of the decision, such as the weight of income volatility and existing debt obligations in the outcome. The explanation is specific to the applicant's case, comprehensible without technical training, and sufficient to decide whether to contest the refusal or correct inaccurate inputs.

How Article 86 Connects to Other Provisions

Article 86 is the individual-facing end of a chain that starts in design: Article 13 transparency and Article 12 logging create the explainability raw material, Article 14 human oversight defines the human role the explanation will describe, Article 26 sets the deployer duties around use and monitoring, and Article 27 embeds complaint arrangements for the deployers it covers. Article 85 provides the complaint route to market surveillance authorities, and Article 99 backs deployer obligations with fines up to 15 million euros or 3 percent of worldwide annual turnover. Like the rest of the high-risk regime, Article 86 applies from August 2, 2026.

Actions to Take Before August 2, 2026

Deployers should inventory their AI-assisted decision points, test whether they could produce a clear and meaningful explanation for a real case today, and fix the gaps, usually missing decision-context capture, before the right becomes enforceable on August 2, 2026. Providers should treat explainability support as a product requirement, because deployers will select systems they can answer for. This article provides general information about Regulation (EU) 2024/1689 and is not advice on any specific decision process.

This article is for informational purposes only and does not constitute legal advice. Regulatory requirements change frequently — verify current rules with official sources. Built by Sawai Gyoseishoshi Office, Hiroshima, Japan.