Risk-based obligations for AI systems, models, providers, deployers, and other actors.
EU AI ACT GOVERNANCE WORLD
Understand the law. Then operate under it.
Move from role and system classification into a persistent governed state: obligations, evidence, gaps, versions, changes, review and revalidation. TA-14 preserves what was established, what remains unsupported, and whether prior reliance still corresponds to the system that exists now.
This workspace does not provide legal advice or certify compliance.
It structures obligations, evidence, gaps, review routes, decisions, records and continuing evidence state.
One governed identity follows each system through classification, evidence, change and revalidation.
Supported propositions remain distinguishable from gaps, stale evidence and review-required conditions.
A changed system or source can trigger revalidation without rewriting the historical record.
ROLE NAVIGATOR
Start Guided Classification Begin with the role you actually hold.
Different actors can carry different obligations for the same system or model. Choose the closest role to open a bounded pathway.
PR
Provider
You develop an AI system or general-purpose AI model and place it on the market or put it into service under your name or trademark.
Open role pathway DEDeployer
You use an AI system under your authority in a professional or organisational context.
Open role pathway IMImporter
You place an AI system bearing the name or trademark of a provider established outside the Union on the Union market.
Open role pathway DIDistributor
You make an AI system available on the Union market without being the provider or importer.
Open role pathway PMProduct Manufacturer
You place an AI system on the market or put it into service together with your product under your own name or trademark.
Open role pathway ARAuthorised Representative
You are established in the Union and hold a written mandate to perform specified tasks on behalf of a provider.
Open role pathway GPGPAI Provider
You develop or place a general-purpose AI model on the market and need a model-level governance pathway.
Open role pathway ?Not Sure?
Use a guided route to identify possible roles, system categories, exceptions, and evidence still needed.
Open role pathway?
GUIDED CLASSIFICATION
Start Guided Classification I do not know whether my system is high-risk—or which role applies.
Walk through a bounded sequence that separates declared facts, unresolved facts, possible classifications, exceptions, and evidence still needed.
REQUIREMENTS EXPLORER
Open the Act by governance problem—not by guesswork.
Each destination preserves the legal source, actor, applicability basis, evidence expectations, unresolved questions, review route, and governed outputs.
01
Risk Classification
Determine whether a system is prohibited, high-risk, transparency-scoped, limited-risk, or outside a claimed category.
Explore requirement02Available
Prohibited Practices
Map use cases against prohibited-practice categories, exceptions, evidence, and escalation boundaries.
Explore requirement03Available
High-Risk AI Systems
Explore lifecycle duties, risk management, data governance, documentation, oversight, logging, and monitoring.
Explore requirement04
General-Purpose AI
Separate model-provider duties, systemic-risk pathways, technical information, copyright policy, and downstream support.
Explore requirement05Available
Article 50 Transparency
Map direct-interaction disclosure, synthetic-content marking, biometric notice, deepfakes, and public-interest text.
Explore requirement06Planned
Conformity Assessment
Preserve the selected assessment route, evidence package, reviewers, findings, corrections, and resulting standing.
View roadmap07Planned
Post-Market Monitoring
Govern continuing performance, incidents, material changes, corrective action, and evidence validity after deployment.
View roadmap08Planned
Incident Reporting
Create bounded routes for detection, classification, chronology, notification, correction, and preserved outcome.
View roadmap09
Human Oversight
Define accountable human authority, intervention capability, escalation, competence, and override boundaries.
Explore requirement10Planned
Technical Documentation
Bind system identity, intended purpose, architecture, data, testing, limits, versions, changes, and evidence ownership.
View roadmap11Available
Fundamental Rights Impact Assessment
Structure affected-person context, risks, safeguards, governance decisions, review, and retained limitations.
Explore requirement12
Recordkeeping and Logs
Preserve identity, chronology, provenance, access, decisions, interventions, changes, and outcomes.
Explore requirementPLATFORM ROADMAP
Planned and expanding modules remain visibly bounded until their dedicated governed workspaces are ready.
GOVERNANCE PROGRESS MAP
Move from an uncertain system description to a verifiable governed record.
Identify
Identify the actor, system, model, product, use case, jurisdiction, and intended purpose.
→Classify
Separate role classification, system category, risk category, and claimed exceptions.
→Determine Applicability
State why each obligation is included, excluded, conditional, or unresolved.
→Map Requirements
Translate applicable requirements into bounded evidence and decision routes.
→Preserve Evidence
Bind claims to documents, tests, owners, versions, chronology, and limitations.
→Review
Challenge reasoning, expose gaps, preserve objections, and correct without erasing history.
→Governed Record
Create a dated, attributable record of applicability, evidence, decisions, and boundaries.
→Independent Verification
Test whether the preserved package still corresponds to the claimed implementation.
The platform must point back to the controlling source.
Official Article
Each requirement route should identify the controlling article and preserve the exact source version used for the determination.
Relevant Recitals
Recitals may provide context but should remain distinguishable from binding operative provisions.
Official Guidance
Commission, AI Office, Board, standards, and authority guidance should be dated, attributed, and never silently substituted for the Regulation.
TA-14 Boundary
TA-14 structures governance routes and evidence. It does not replace the official text, competent legal advice, conformity assessment, or regulator judgment.
Do not collapse the law, the evidence, and the conclusion into one layer.
A regulation states obligations. An organisation declares how those obligations apply. Evidence supports that declaration. Review tests the evidence and reasoning.
01Obligation02Applicability03Evidence04Determination05Review06Record
ARTICLE 50 TRANSPARENCY WORKSPACE
Open dedicated Article 50 workspace Map the obligation before claiming the outcome.
These demonstration pathways separate provider and deployer duties, identify evidence dependencies, expose missing proof, and define review-ready outputs.
Role
Article 50(1)READY TO MAP
TA-14-EU-AIA-50-1Direct AI interaction disclosure
Provider
Map whether a system interacts directly with natural persons and whether the disclosure design clearly informs them that they are interacting with AI.
Potential scope
- Interactive AI systems
- Chatbots and conversational systems
- Direct user-facing AI interfaces
Evidence to preserve
- Intended-purpose statement
- User-interface captures
- Disclosure timing and wording
- Exception analysis
- Version and deployment record
Governed outputs
- Applicability record
- Disclosure evidence map
- Gap and exception record
- Review-ready route package
Article 50(2)EVIDENCE GAP
TA-14-EU-AIA-50-2Machine-readable marking and detectability
Provider
Preserve how synthetic or manipulated audio, image, video, and text outputs are marked in a machine-readable format and made detectable where technically feasible.
Potential scope
- Generative AI systems
- Synthetic media systems
- AI-generated or manipulated content
Evidence to preserve
- Marking architecture
- Detection testing
- Interoperability evidence
- Robustness evidence
- Technical-feasibility determination
Governed outputs
- Marking implementation record
- Detectability test record
- Technical limitation record
- Provider evidence package
Article 50(3)REVIEW REQUIRED
TA-14-EU-AIA-50-3Emotion recognition and biometric categorisation notice
Provider or Deployer
Map notice obligations where natural persons are exposed to emotion-recognition or biometric-categorisation systems, subject to applicable exceptions.
Potential scope
- Emotion-recognition systems
- Biometric-categorisation systems
- Workplace or public-facing deployments
Evidence to preserve
- System classification
- Deployment context
- Notice design
- Exception analysis
- Affected-person pathway
Governed outputs
- Applicability determination
- Notice implementation record
- Exception and limitation record
- Deployment evidence package
Article 50(4)READY TO MAP
TA-14-EU-AIA-50-4Deepfake and public-interest text disclosure
Deployer
Map deployer-side disclosure for deepfakes and certain AI-generated or manipulated text published to inform the public on matters of public interest.
Potential scope
- Deepfakes
- AI-generated public-interest text
- Professionally deployed synthetic media
Evidence to preserve
- Content classification
- Editorial-control record
- Disclosure wording and placement
- Publication chronology
- Exception analysis
Governed outputs
- Deployer disclosure record
- Editorial-control determination
- Publication evidence package
- Exception and boundary record
GOVERNED RECORD ARCHITECTURE
Open Governed Records Every material claim should become a dated, attributable record.
01
Applicability Record
Preserves actor, role, system, use case, jurisdiction, scope, exceptions, and the basis for inclusion or exclusion.
Create record 02Evidence Map
Binds each obligation or commitment to documents, technical artifacts, tests, owners, versions, and unresolved gaps.
Create record 03Transparency Implementation Record
Preserves marking, notice, disclosure, placement, timing, detectability, and deployment state.
Create record 04Independent Review Record
Preserves reviewer identity, scope, evidence reviewed, findings, objections, corrections, and limitations.
Create record 05Change Record
Shows what changed, when, why, by whom, and whether prior evidence remains valid.
Create record 06Outcome Record
Preserves what was approved, held, denied, escalated, superseded, withdrawn, or left unresolved.
Create recordCONNECTED EXCHANGE PATHWAYS
The EU AI Act world is an operating layer inside the institution—not an isolated compliance page.
01
EU AI Act Command Center
Operate persistent system passports, obligation states, evidence gaps, change pressure and revalidation.
Open pathway 02AI Governance Registry
Preserve governance identity, establishment date, versions, claims, non-claims, evidence, and stewardship.
Open pathway 03Governance Routes
Convert obligations into inspectable pathways with inputs, gates, evidence, outputs, and explicit failure states.
Open pathway 04Governed Records
Preserve applicability, evidence, review, disclosure, change, and outcome records.
Open pathway 05Independent Review
Discover professionals through declared expertise, evidence signals, artifacts, and visible limitations.
Open pathwayKnow what applies. Know why. Know what evidence exists. Know what changed.
Build a persistent system passport, map obligations, preserve evidence, expose gaps, detect change and revalidate prior reliance before the representation outruns the condition underneath it.