Praxikon

Independent AI Act implementation knowledge

From official source to demonstrable action

Ask your question and see which AI Act rule applies, what materially changed and which evidence you should retain. With legal status, practical analysis and a direct path to the official source.

What do you need clarity on?

Try:Our chatbot talks to customersWe screen CVs with AIEmployees use ChatGPTAll direct answers

Open access. No account required. Monitor updates show their official source, legal status and latest review.

Praxikon provides source-traceable EU AI Act knowledge for professionals, organisations and AI agents.

In force now

Shifted (Digital Omnibus)

The delay is runway for register and evidence, not a pause.

Next date

2 December 2026

End of the transition for machine-readable marking (Art. 50(2)) and new prohibitions on non-consensual synthetic content.

Track changes in the Monitor

AI Act Implementation Map

One system. Five routes. One usable record.

Move from system context to applicability, first actions, evidence and official sources in a few minutes. The map remains public and account-free.

Build your implementation map

No account. No email. No personal data required.

  1. 01Describe one systemFixed answer choices only. No name, account or email.
  2. 02See what affects youArticle 4, Article 50 by paragraph, Annex III, FRIA and GPAI in one map.
  3. 03Take a record with youActions, evidence, official sources, versions and reassessment as JSON or PDF.

AI Act Monitor

What is materially changing now

Not an unedited news stream, but verified changes that affect obligations, planning or evidence.

Last editorial review:

View all changesRead the methodology
  2. ApplicableAction required

    Enforcement of GPAI obligations has started

    What changed

    Since 2 August 2026, the Commission can enforce the obligations for providers of general-purpose AI models. The substantive obligations have applied to new models since 2 August 2025.

    First next step

    Record whether the organisation is a general-purpose AI provider, a downstream provider or only a deployer.

    Impact and evidenceOfficial source

Obligation to evidence

Five core routes to a defensible record

Do not start with an isolated document. First determine the applicable obligation, take the appropriate action and then retain the evidence.

  1. Article 4AI literacy and evidence of measuresQuestion to answerWhich role-, context- and risk-based measures are appropriate?Evidence to retainRole matrix, AI inventory, measures log and periodic review.
  2. Article 50Transparency for AI interaction and contentQuestion to answerWhich disclosure, marking or visible information is required?Evidence to retainApplicability decision, disclosure text, owner and review evidence.
  3. Annex IIIHigh-risk classificationQuestion to answerDoes the use fall within a high-risk domain and does the Article 6(3) filter apply?Evidence to retainClassification rationale, filter analysis, registration and reassessment trigger.
  4. Article 27Fundamental rights impact assessmentQuestion to answerIs a FRIA required and how should it connect to the DPIA?Evidence to retainFRIA report, affected-person impact, measures and decision record.
  5. Articles 51-55GPAI and systemic-risk modelsQuestion to answerWhat is your role around the model and which additional duties follow?Evidence to retainTechnical information, copyright policy, downstream information and risk file.
Not sure which route applies?View public templates

Explore the analysis

Open specialist source routes when the concise obligation card does not provide enough detail.

Execution

All knowledge here is free. You do not have to do the execution alone.

Embed AI

Governance, register and FRIA arranged

For the organisation that wants to turn obligations into a working whole: scope, AI register, classification, FRIA and the evidence file, guided in a fixed approach.

See the Embed AI approach

LearnWize

AI literacy demonstrable per role

For the people behind the systems: role-based learning paths and assessment under Article 4, with an audit-ready file per employee.

See LearnWize

Editorial methodology

Trust should be verifiable

Our value is not more text, but traceable selection, visible freshness and a consistent path from source to action.

  1. Official sources first

    Every material change links to at least one canonical source, such as EUR-Lex, the European Commission or the AI Office.

  2. Facts and analysis remain separate

    We distinguish what the source establishes from our editorial analysis and possible next steps.

  3. Human review before publication

    Automated checks help identify changes. Publication follows only after source, status and materiality have been reviewed.

  4. Open and verifiable

    Reading, searching, checking sources and using core routes requires no account. Review dates and correction routes remain visible.

Read the full methodologyReport a correction or missing source

Newsletter

Every Tuesday, the AI Act week ahead in 5 minutes

A practical briefing on deadlines, new guidance and enforcement, so you know what matters this week. No spam and you can unsubscribe in one click.

Practical and short · No spam · One-click unsubscribe