Administration of Justice & Democratic Processes , Annex III Point 8 High-Risk AI
AI used by a judicial authority, or on its behalf, in researching and interpreting facts and the law, and in applying the law to a concrete set of facts, or used for influencing the outcome of an election or referendum or the voting behaviour of natural persons, is high-risk under Annex III Point 8.
Scope, Definitions and Boundary with Prohibited Practices
Annex III Point 8 is split into two sub-points: 8(a) covers AI used by or on behalf of judicial authorities in legal research, fact interpretation, and application of law to facts. 8(b) covers AI used to influence the outcome of elections or referenda or the voting behaviour of natural persons, with the important carve-out that systems whose output natural persons are not directly exposed to (e.g. backend administrative tools used by political parties for logistics) are excluded.
Point 8 is unusual because it straddles two very different sectors, courts and elections, that share a common democratic-integrity concern. Obligations under the DSA (Digital Services Act) relating to systemic risks on very large online platforms and search engines intersect significantly with Point 8(b) for platform-deployed political-influence AI.
Example Systems Under Annex III Point 8
Illustrative examples, each Member State's market surveillance authority may refine classification guidance over time.
Example
AI-assisted judicial decision-drafting tool
Point 8(a) high-risk. Judicial independence concerns mean human oversight obligations are particularly salient.
Example
Predictive-case-outcome platform used by judges
Point 8(a) high-risk. Scientific and procedural validity must be documented under Article 9.
Example
Political microtargeting engine for voter outreach
Point 8(b) high-risk where natural persons are directly exposed to the AI output.
Example
Back-office campaign logistics AI
Out of Point 8(b) scope, carve-out for ancillary administrative activities with no direct exposure.
Example
Generative-AI content labelling for political ads
Triggers both Article 50 transparency (labelling) and, depending on function, Point 8(b) high-risk.
The Articles That Apply
High-risk classification under Annex III triggers the full Chapter III, Section 2 obligation set. Deployers pick up additional obligations under Chapter III, Section 4.
| Article | Obligation | What It Means in Practice |
|---|---|---|
| Art. 9 | Risk Management System | Point 8(a) risk management must address scientific validity of legal-AI outputs and the risk of automation bias in judicial decision-making. Point 8(b) must address democratic-process integrity. |
| Art. 10 | Data Governance | Training data from judicial corpora may encode prior discriminatory outcomes. For 8(b), political-preference data is special-category data under GDPR. |
| Art. 13 | Transparency | For 8(b), transparency duties intersect with DSA Article 26 (online advertising transparency) and with Regulation (EU) 2024/900 on the transparency and targeting of political advertising. |
| Art. 14 | Human Oversight | Judicial independence means human oversight for 8(a) must preserve the judge's genuine discretion, not merely formal review of AI output. |
| Art. 26 | Deployer Obligations | Public-authority deployers (courts, electoral management bodies) subject to Article 27 FRIA. Private political-campaign deployers of 8(b) systems are not automatically FRIA-bound unless deemed to provide a public service. |
| Art. 27 | FRIA | Mandatory for public-authority deployers. For 8(b), recommended even where not mandatory given the systemic-risk posture. |
| Art. 49 | EU Database Registration | Standard. Judicial-authority systems may be registered in the non-public section. |
| DSA, Political Advertising Reg. | Sectoral Overlap | Regulation (EU) 2024/900 imposes transparency and targeting restrictions on political advertising. DSA obligations on VLOPs/VLOSEs apply independently. |
Pitfalls to Avoid on Annex III Point 8
Recurrent patterns seen in pre-enforcement readiness assessments.
Treating judicial-research AI as low-risk because it is 'only research'
The plain text of Point 8(a) covers AI used in researching and interpreting facts and law. Research functions are in scope regardless of whether they directly produce a decision.
Exempting backend political-campaign AI
The carve-out for 'administrative or logistical activities' is narrow. Anything affecting voter exposure, even indirectly, falls within Point 8(b).
Missing the Political Advertising Regulation overlap
Regulation (EU) 2024/900 entered into force in 2024 with staggered application. It regulates targeting and transparency of political advertising independently of the AI Act.
Generative-AI content with no labelling
Article 50(4) requires disclosure where AI-generated or manipulated content 'appreciably resembles' real persons, objects, places, or events. Political deepfakes are the canonical case.
Operationalise Point 8 Compliance
The Full Readiness Bundle gives your legal and compliance teams the 58-Point Compliance Checklist, Annex III Classification Matrix, FRIA template, Annex IV Technical Documentation Checklist, and 7 more documents, all aligned to the Annex III high-risk regime (now applying from 2 December 2027 under the May 2026 Omnibus deal).
one-time · instant download
Get the Full Bundle →Also available: Checklist Pack $149 · White-Label $999/yr
Next Steps
Deadline Planning
15-week compliance work plan for the EU AI Act's phased timeline.
Week-by-Week Plan →