AI used in courts, legally effective dispute resolution, elections and referenda can sit directly inside Annex III high-risk pathways. The central governance question is whether the system merely supports an accountable human process—or materially influences a legal or democratic outcome without a durable evidence and authority record.
The Regulation classifies specified AI used in the administration of justice and democratic processes as high-risk, including certain judicial-assistance systems and systems intended to influence election or referendum outcomes or voting behaviour. It also states that final judicial decision-making must remain human-driven and distinguishes purely ancillary administrative activities.
Start with the actual system, role, use case and evidence boundary. The same regulation can produce different obligations for different actors and systems.
Annex III classifies certain AI intended to assist judicial authorities, or act on their behalf, in researching and interpreting facts and law and applying the law to a concrete set of facts as high-risk.
AI used by alternative dispute resolution bodies can also fall on the high-risk route when the system performs those justice-related functions and the proceeding produces legal effects for the parties.
Annex III also treats AI intended to influence the outcome of an election or referendum, or the voting behaviour of natural persons in the exercise of their vote, as high-risk, subject to the Regulation’s exception for purely administrative or logistical campaign tools whose outputs are not directly exposed to voters.
Preserve the system’s role, authority boundary, source material, model and version, human review, reasoning inputs, override path, logs, material changes and the evidence showing that AI supported rather than silently replaced the accountable human decision.
The goal is not merely to reach an answer. It is to preserve what facts, evidence, scope and limitations supported that answer at that time.
Separate legal research, fact analysis, law application, administrative support, dispute resolution, campaign logistics, voter-facing influence and election optimisation. These functions do not all sit on the same AI Act route.
Determine whether the system materially influences a listed justice or democratic-process use. If a provider relies on the Article 6(3) non-high-risk route, preserve the documented basis before placing the system on the market or putting it into service.
Connect source material, system output, human interpretation, decision authority, review, logging, limitations, challenge or appeal pathways and any correction record so the outcome remains reconstructable.
A new model, changed legal source, jurisdiction, workflow, dispute-resolution role, voter-facing feature or campaign use can make the prior classification and evidence stale.
Choose the smallest operating tier that fits the portfolio today. Upgrade when system count, team size or governance scope actually requires it.
Keep a living system-level evidence record with obligations, gaps, versions and revalidation state.
START EVIDENCE PASSPORT →Coordinate evidence, owners, documentation, incidents and team compliance work in one governed workspace.
START COMPLIANCE WORKSPACE →Operate broader high-risk, GPAI, FRIA, post-market and material-change governance across a growing portfolio.
START GOVERNANCE PRO →Run institutional governance with expanded users, authority workflows, examiner rooms and portfolio reporting.
START INSTITUTION →Use the free classifier to establish the system, intended purpose, possible actor role, EU exposure and unresolved facts. When continuing evidence infrastructure is needed, paid access begins at $19 per month. Independent human readiness review remains a separate service.
Certain judicial AI systems are. Annex III covers AI intended to assist a judicial authority, or act on its behalf, in researching and interpreting facts and the law and applying the law to a concrete set of facts.
The Regulation’s recitals state that AI may support judges and judicial independence, but should not replace the final human decision. Final decision-making must remain human-driven.
No. Purely ancillary administrative activities that do not affect the actual administration of justice in individual cases—such as anonymisation, pseudonymisation, personnel communication or routine administrative tasks—are distinguished from the high-risk judicial route.
Certain AI intended to influence election or referendum outcomes or the voting behaviour of natural persons is high-risk. The Regulation excludes purely administrative or logistical campaign tools whose outputs are not directly exposed to natural persons.
Preserve system identity and version, intended purpose, authority, source material, classification basis, human decision role, logs, limitations, appeal or challenge routes, changes and the evidence supporting each material decision or influence claim.
No. TA-14 can preserve the classification basis, authority claim, evidence, decision scope, gaps, review history and revalidation state. It does not provide legal advice, judicial authorisation, conformity assessment, certification or regulatory approval.
TA-14 Exchange Activity
Live cumulative activity recorded across the public Exchange surface.
···
Visitors
Recorded public visitors
···
Page Views
Recorded Exchange views