EU AI Act Compliance
for AI Product Teams
Classify each AI system into the correct risk tier and generate the exact documentation the EU AI Act asks for — automatically.
The EU AI Actis the first comprehensive regulation of artificial intelligence. It classifies AI systems into four risk tiers and imposes graded obligations from "banned outright" to "transparency only". If your product uses AI and any of your users touch the EU market, the Act applies to you.
The four risk tiers
Social scoring, real-time biometric identification in public spaces (with narrow exceptions), manipulative or exploitative systems targeting vulnerable groups. These practices are banned outright.
AI used in employment, credit scoring, education, critical infrastructure, medical devices, law enforcement, and migration. Requires full risk management, technical documentation, logging, human oversight, and a conformity assessment.
Chatbots, emotion recognition, generative AI content. Transparency duties only — users must be told they are interacting with AI, and synthetic content must be labelled.
Spam filters, recommender systems, game AI. No obligations under the Act, but voluntary codes of conduct are encouraged.
Deadlines
The Act came into force in 2024 and phases in progressively. The first 2026 deadline hits prohibited practices, followed by general-purpose AI model obligations. High-risk system obligations become enforceable in the following window. Missing the deadline for a high-risk system exposes the provider to fines of up to 7% of global annual turnover.
Zilonex Govern tracks each phase against your registered models so you always know which of your systems are already in scope and which are on the horizon.
Documentation a high-risk system needs
- Risk management system covering the full lifecycle
- Data governance and training-data documentation
- Technical documentation (Article 11 / Annex IV)
- Automatic event logs during operation
- Transparency and instructions for deployers
- Human oversight measures
- Accuracy, robustness, and cybersecurity evidence
- Post-market monitoring plan
- Conformity assessment and CE-style declaration
How Zilonex Govern automates the paperwork
Register a model, answer a short intent questionnaire, and Zilonex Govern classifies it, opens the correct set of controls, links your risk register entries, and generates the Article 11 technical documentation as a signed PDF.
When a model version changes, the classifier re-runs and any impacted controls re-open automatically. No more "is our AI Act pack still current?" standups.
Frequently asked questions
What is the EU AI Act?
The EU AI Act is the first comprehensive law regulating artificial intelligence. It applies to any provider or deployer of an AI system that is placed on the market or whose output is used, and it classifies AI systems into four risk tiers with different obligations for each tier.
What are the four risk tiers?
The four tiers are: prohibited (social scoring, real-time biometric identification in public spaces except narrow cases, manipulative systems), high-risk (AI in employment, credit, education, critical infrastructure, medical devices, law enforcement), limited-risk (chatbots and generative AI with transparency duties), and minimal-risk (spam filters, game AI — no obligations).
When are the deadlines?
The Act came into force in 2024 and phases in progressively. Prohibited practices hit their 2026 deadline first, followed by general-purpose AI model obligations, then the full high-risk regime becomes enforceable in the months after. Zilonex Govern tracks each deadline against your registered models automatically.
What documentation does a high-risk AI system need?
A high-risk system needs a risk management system, data governance documentation, technical documentation, automatic event logs, transparency information for deployers, human oversight measures, accuracy/robustness/cybersecurity evidence, and a conformity assessment. Zilonex Govern produces this pack from your model registry entries.
How does Zilonex Govern classify AI systems?
You answer a short intent questionnaire per model — use-case, deployment context, output type, decision autonomy. The classifier maps the answers onto the Act's Annex III list and returns the risk tier with a reasoning trail you can review and export.
Is Zilonex Govern a substitute for legal advice?
No. Zilonex Govern is a tooling and evidence platform. It captures the technical facts the Act asks for and generates the documentation pack, but final legal sign-off on classification and conformity should always involve qualified counsel.
Classify your first AI system
Automated risk-tier classification and Article 11 documentation from $99/month.
Sign up for Zilonex Govern