Your Employee Just Invoked Their Right to Know How Your AI Tool Assessed Them: Answering the Article 86 Individual Transparency Questions
It happened during due diligence, not after deployment.
A 4,200-person German logistics firm is two weeks from signing a €190K ARR contract for your AI workforce management platform. Then their works council chair emails the procurement lead with a two-line message: "Before we proceed, we require answers to how this system will explain individual assessments to employees who request it under Article 86."
The procurement lead forwards it to you. You have four business days to respond.
Article 86 is the EU AI Act question most HR tech CTOs haven't prepared for — because it's not about your architecture or your training data. It's about what happens after your tool makes a decision and the person that decision affects asks: why?
What Article 86 Actually Requires
Article 86 of the EU AI Act gives individuals who are subject to a decision made by a high-risk AI system the right to request an explanation of the role the AI played in that decision and the main parameters considered.
This matters for HR tech for one specific reason: Annex III, Category 4 classifies AI systems used for "recruitment or selection of natural persons, notably for advertising vacancies, screening or filtering applications, evaluating candidates in the course of interviews or tests" and "for making decisions on promotion and termination of work-related contractual relationships" as high-risk. If your product influences any of these decisions — directly or as an input — Article 86 applies.
The explanation obligation sits on the deployer (typically your enterprise customer), but the deployer can only fulfil it if you, the provider, have built the system to support it. When a works council asks "can your tool explain its assessments to individual employees?", they're asking whether you've built that capability in. If you haven't, the deal is at risk regardless of how good your product is.
The Five Questions You'll Get in This Section
1. "What information does an employee receive if they request an Article 86 explanation?"
What they're asking: If an employee subject to an AI-assisted performance evaluation or shortlisting decision invokes their Article 86 right, what does the explanation actually look like? Is it a human-readable narrative, a list of features, a confidence score, a comparison to peer cohorts?
How to answer: Be concrete. An explanation that says "you received a lower ranking because the model weighted collaboration signals below the median" is a better answer than "the AI considered multiple factors." Describe the format of the explanation output (structured text, summary table, feature breakdown), who receives it (employee directly, via HR business partner, or via manager), and whether it's generated on demand or logged at decision time.
If your product doesn't currently produce employee-facing explanations, say so — and describe the roadmap. Procurement teams score you on honesty and direction of travel, not just current state.
2. "Which AI features or parameters are disclosed in the explanation?"
What they're asking: This is the Annex IV Article 13(3)(b) question in disguise. Buyers need to know whether the explanation exposes the parameters that actually drove the decision or just surface-level outputs.
How to answer: List the categories of information your explanation covers. For a performance assessment tool, this might include: weighted inputs to the overall score (meeting participation rate, task completion rate, peer review sentiment score), confidence interval on the overall score, any flags raised by the system (e.g., "low data volume warning — fewer than 30 assessed data points"), and date range of data considered.
Do not claim the explanation reveals proprietary model weights if it doesn't. Courts in France, Germany, and the Netherlands have already ruled that "black box score with no path to challenge" fails the purpose of transparency obligations. A good procurement team will probe this.
3. "How does an employee challenge an AI-assisted decision under your system?"
What they're asking: Article 86 is linked to the broader EU AI Act principle of meaningful human oversight (Article 14). If an employee can request an explanation but has no channel to contest a decision, the explanation is cosmetic.
How to answer: Describe the override or challenge path in your product. Who does the employee address the challenge to? What workflow does that trigger in your product? Is the challenge logged and timestamped? Does a human reviewer have to formally document their decision after a challenge?
The answer does not have to be elaborate. Even a simple workflow — "employee submits a correction request via HR portal → HR business partner reviews the AI score and underlying inputs → outcome logged with rationale" — is defensible. What is not defensible is "employees can talk to their manager" with no system-level record.
4. "How long are explanation records retained?"
What they're asking: Article 12 of the EU AI Act requires automatic event logging for high-risk AI systems. Article 86 explanations are a subset of those events. Buyers need to know that explanation records exist and are recoverable for audit.
How to answer: State your log retention period for AI-generated assessments and their corresponding explanation records. If your standard is 36 months, say so. If the deployer can configure a longer retention period, note that. Confirm that explanation records include: the employee identifier (anonymised or pseudonymised per your data governance), the assessment date, the parameters surfaced, and the version of the model that produced the assessment.
5. "Does your system support the right not to be subject to solely automated decisions under Article 22 GDPR?"
What they're asking: This one spans both the EU AI Act and GDPR. Article 22 GDPR prohibits automated-only decisions with legal or similarly significant effects unless the individual has explicitly consented or the decision is authorised by law with suitable safeguards. Article 86 EU AI Act reinforces this by requiring the explanation to describe what role the AI played.
How to answer: The key is the phrase "solely automated." If a human decision-maker reviews the AI output before any employment-affecting action is taken — shortlisting a candidate, triggering a performance review conversation, initiating a role reassignment — your product supports Article 22 compliance. State exactly what human review your product requires (configurable, mandatory, recommended) and whether the system enforces it (e.g., the deal cannot be progressed in your platform without a human acknowledgement step).
What Complizo Does With Article 86 Questions
Complizo maps Article 86 questions to your actual product features — not to a generic "our system supports transparency" paragraph. When a works council sends a questionnaire section on individual transparency rights, Complizo surfaces the answers you've already given to related questions (your logging setup, your human override workflow, your data retention policy) and composes a consistent Article 86 response anchored to the specific capabilities your product ships with.
The result is an answer that cites your feature, not a legal abstract. That's the difference between a procurement team that says "thanks, we'll proceed" and one that sends a follow-up with six more questions.
Try Complizo free at complizo.com