How AI Explained is made
How AI Unpacked separates explanation, measurement, rules and predictions.
Purpose: to explain artificial intelligence in plain language — not to persuade readers that AI is either a miracle or a threat.
Principles
- Explanation before jargon. First we say what something means in everyday life. The technical term comes after.
- A fact is not a forecast. We clearly separate what has been measured, what comes from research and what is a scenario.
- Sources are visible. Priority goes to the official texts of laws, international expert bodies and direct research sources.
- Checks carry a date. We regularly review important claims, especially on law, safety, market comparisons and model capabilities.
- AI is not the editor. A responsible person approves texts, images, corrections and publication.
- No hidden advertising. An explanation is not a recommendation of any brand or provider.
- No unfounded claims. Where there is no consensus, we present the uncertainty and the criteria for checking.
- Accessibility. Short paragraphs, descriptive links, text alternatives for images, and meaning that does not rely on colour.
Reliability labels
| Label | Meaning |
|---|---|
| Basic explanation | An explanation of an established concept. |
| Current figure | A number or fact that needs a date and a source. |
| Legal framework | A summary of a law; check the text in force before making decisions. |
| Open question | An expert debate or an unanswered question. |
| Scenario | A possibility, not a prediction. |
Review schedule
- Every month: dates of legal rules; AI agents and safety; corrections to sources.
- Every three months: claims about progress, achievements and the market.
- Every six months: basic definitions, the map of the field and the glossary.
- After a major change: the article is corrected, a new version is issued and the change is recorded.
Selected primary and reference sources
- OECD — updated definition of AI
- Stanford HAI — AI Index Report 2026
- NIST — AI Risk Management Framework
- NIST — Generative AI risk profile
- European Commission — AI Act
- European Commission — AI Act: questions and answers
- European Commission — transparency under Article 50
- European Commission — AI literacy under Article 4
- OWASP — Top 10 for LLMs, 2026 edition
- C2PA — standards for tracing digital content
- IEA — Key Questions on Energy and AI (2026)
- UNESCO — Recommendation on the Ethics of AI
- WHO — ethics and governance of multimodal models in health
- EUR-Lex — Regulation (EU) 2024/1689, as amended
Editorial note
This material is educational. It is not a substitute for medical, legal or other professional advice. The dates on which rules apply were checked on 1 November 2026, and we update them whenever the legal framework changes.
Last reviewed 1 November 2026. An explainer, not professional, legal or medical advice. For rules and figures that change, check the official version.