EU AI ACT GOVERNANCE WORKSPACE

Turn regulatory obligations into inspectable evidence routes.

The EU AI Act does not become governable merely because an organisation has a policy, checklist, legal memo, or dashboard. TA-14 separates applicability, evidence, determination, review, commitment, execution, and preserved outcome so each claimed compliance pathway can be inspected under scrutiny.

RegulationEU 2024/1689

Risk-based obligations for AI systems, models, providers, deployers, and other actors.

Current focusArticle 50

Interaction disclosure, marking, detectability, deepfakes, and public-interest text.

Code statusVOLUNTARY

Signing can support a compliance pathway but is not conclusive proof of compliance.

Alternative routePERMITTED

Non-signatories remain responsible for demonstrating adequate alternative measures.

ROLE NAVIGATOR

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.

Start Guided Classification
?
GUIDED CLASSIFICATION

I do not know whether my system is high-risk—or which role applies.

Walk through a bounded sequence of questions that separates declared facts, unresolved facts, possible classifications, exceptions, and evidence still needed.

Start Guided Classification
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.

01Expanding

Risk Classification

Determine whether a system is prohibited, high-risk, transparency-scoped, limited-risk, or outside a claimed category.

Explore requirement
02Available

Prohibited Practices

Map use cases against prohibited-practice categories, exceptions, evidence, and escalation boundaries.

Explore requirement
03Available

High-Risk AI Systems

Explore lifecycle duties, risk management, data governance, documentation, oversight, logging, and monitoring.

Explore requirement
04Expanding

General-Purpose AI

Separate model-provider duties, systemic-risk pathways, technical information, copyright policy, and downstream support.

Explore requirement
05Available

Article 50 Transparency

Map direct-interaction disclosure, synthetic-content marking, biometric notice, deepfakes, and public-interest text.

Explore requirement
06Planned

Conformity Assessment

Preserve the selected assessment route, evidence package, reviewers, findings, corrections, and resulting standing.

View roadmap
07Planned

Post-Market Monitoring

Govern continuing performance, incidents, material changes, corrective action, and evidence validity after deployment.

View roadmap
08Planned

Incident Reporting

Create bounded routes for detection, classification, chronology, notification, correction, and preserved outcome.

View roadmap
09Expanding

Human Oversight

Define accountable human authority, intervention capability, escalation, competence, and override boundaries.

Explore requirement
10Planned

Technical Documentation

Bind system identity, intended purpose, architecture, data, testing, limits, versions, changes, and evidence ownership.

View roadmap
11Available

Fundamental Rights Impact Assessment

Structure affected-person context, risks, safeguards, governance decisions, review, and retained limitations.

Explore requirement
12Expanding

Recordkeeping and Logs

Preserve identity, chronology, provenance, access, decisions, interventions, changes, and outcomes.

Explore requirement
PLATFORM ROADMAP

Planned and expanding modules return here rather than sending visitors to unfinished destinations. Available modules open their dedicated workspaces.

GOVERNANCE PROGRESS MAP

Move from an uncertain system description to a verifiable governed record.

01

Identify

Identify the actor, system, model, product, use case, jurisdiction, and intended purpose.

02

Classify

Separate role classification, system category, risk category, and claimed exceptions.

03

Determine Applicability

State why each obligation is included, excluded, conditional, or unresolved.

04

Map Requirements

Translate applicable requirements into bounded evidence and decision routes.

05

Preserve Evidence

Bind claims to documents, tests, owners, versions, chronology, and limitations.

06

Review

Challenge reasoning, expose gaps, preserve objections, and correct without erasing history.

07

Governed Record

Create a dated, attributable record of applicability, evidence, decisions, and boundaries.

08

Independent Verification

Test whether the preserved package still corresponds to the claimed implementation.

SOURCE-BOUND GOVERNANCE

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.

THE TA-14 DIFFERENCE

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

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.

Jump to Article 50 Pathways
Role
4 pathwaysNo backend evidence submission or legal determination is connected yet.
Article 50(1)READY TO MAP
TA-14-EU-AIA-50-1

Direct 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-50-2

Machine-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-50-3

Emotion 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-50-4

Deepfake 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
CODE OF PRACTICE PATHWAY

Signing the Code and complying with Article 50 are not the same claim.

The Code of Practice is a voluntary tool for demonstrating compliance with specified transparency obligations. Adherence can create a more predictable route, but it is not conclusive evidence of compliance.

SIGNATORY ROUTE

Adhere to the Code

  • Identify the exact commitments that apply.
  • Bind implementation evidence to each commitment.
  • Preserve testing, exceptions, limitations, and updates.
  • Demonstrate continuing adherence rather than one-time signature.
Open signatory route
ALTERNATIVE ROUTE

Use other adequate measures

  • Document why the selected measures satisfy Article 50.
  • Map differences from the Code and justify those differences.
  • Prepare evidence for competent-authority information requests.
  • Preserve gap analyses, testing, and technical limitations.
Open alternative route
GOVERNED RECORD ARCHITECTURE

Every material claim should become a dated, attributable record.

Open Governed Records
CONNECTED EXCHANGE PATHWAYS

The EU AI Act workspace should not become another isolated compliance page.

NO ADMISSIBLE EVIDENCE. NO ADMISSIBLE EXECUTION.

The deadline may start the obligation. It does not complete the evidence chain.

The TA-14 EU AI Act workspace is designed to make each regulatory pathway inspectable: what applies, why it applies, what evidence exists, what remains missing, who reviewed it, what changed, and what decision the evidence can actually support.

TA-14 Exchange Activity

Public activity recorded across the Exchange

Visitors

Page Views