Serious game · Education · AI governance
A serious game about the EU AI Act
The AI Act is a technical text describing deeply human decisions: who gets hired, who can access a service, who ends up under surveillance. NO AI ACT turns those decisions into a playable investigation. This page explains why.
01Why a game, and not a lecture
Reading that "risk depends on the context of use" is easy to forget. Misclassifying a system because you skipped the data table — and watching a neighbourhood slide to the bottom of every list — is hard to forget.
A serious game puts the learner inside the mechanism: not a spectator of a regulation, but a decision-maker who must apply it with incomplete information, vendors that minimise and citizens who file complaints. It is the same reason simulations have been used for decades in medical, legal and management training.
02Four design choices
Risk-based thinking
The game follows the AI Act's logic: almost nothing is banned outright. Purpose, data used, context and effects on rights are what count. The ambiguous cases exist on purpose.
Evidence-based reasoning
Every decision must be grounded on cited exhibits. A correct classification built on weak evidence is open to challenge: as in reality, the act has to stand.
Visible human impact
Consequences are measured on the city's indicators — rights, trust, social control — and in concrete stories: the mother who loses nursery priority, the neighbourhood that never learns why it is last.
Reflective debriefing
After each report, a panel explains the reasoning: what to observe, which rule connects evidence and decision, how to reason next time. Mistakes become teaching material.
03Serious games and regulatory literacy
Complex regulation has a specific teaching problem: its consequences are far away from the people studying it. An article about human oversight stays abstract until you have watched a "human check" fail because it existed only on paper.
- Safe failure — in a simulation you can misclassify a system without real harm, and the mistake produces the most durable lesson.
- Intentional ambiguity — real regulatory skill is not recalling the article but qualifying a grey case: the game is built around grey cases.
- Compressed consequences — effects that take years to surface in reality (distrust, exclusion, opacity) become visible in minutes on the city's indicators.
- Vocabulary in context — deployer, GPAI, high-risk, contestability: the terms stick because you need them to decide, not to pass a test.
The same approach also fits broader AI literacy and digital citizenship programmes, where the AI Act provides the structure but the skill being trained is reading any automated system critically.
04What you learn, concretely
The 13 cases cover the regulation's load-bearing categories, always starting from concrete situations:
- Prohibited practices — social scoring and its boundaries: why one score is banned and another, similar one is not.
- High-risk systems — healthcare, education, employment, welfare: obligations, documentation, effective human oversight.
- Transparency obligations — synthetic content and chatbots: when people have the right to know they are talking to a machine.
- GPAI and downstream use — what changes when a general-purpose model enters decisions that matter.
- Responsibility — provider, deployer or authority: who carries the obligations, and why a disclaimer is not enough.
The step-by-step journey is described in how it works; the game does not cover the entire regulation and is not a compliance course: it is a reasoned introduction to its core concepts.
05Why it is called "NO AI ACT"
The title describes the narrative premise, not a political position: the game takes place in a city where the AI Act's safeguards do not yet exist. By playing, you see what happens without them — and that is how the regulation stops feeling like abstract bureaucracy.
NO AI ACT is not a campaign against the AI Act. It is an educational simulation of an unregulated society, designed to show why AI regulation matters.
06What the game is NOT
- Not legal advice and not an official source: it is a simplified educational version of Regulation (EU) 2024/1689.
- Not a compliance or certification tool.
- Not an empirically validated product: external playtesting is ongoing and the project makes no measured-effectiveness claims.
- Not a quiz: there is no answer list to memorise.
07Try it
The fastest way to understand the project is to open a case file: one case takes 10–15 minutes, in the browser, no registration.
To go deeper into the method and the subject: serious games for AI regulation (the methodology page), the AI Act risk categories and the plain-language guide.