NO AI ACT.

EUROPE · 2032 · ALGORITHMIC INCIDENT UNIT

The AI Act never entered into force.Now investigate the consequences.

Thirteen case files. Artificial intelligence systems. Real people. Examine the evidence, classify the risk and sign your inspection report.

SERIOUS GAMEFREEOPEN SOURCEIT / ENNO ACCOUNT

NO AI ACT is not a campaign against the AI Act: it is an educational simulation of an unregulated society, where the regulation never entered into force.

CASE FILE 01 / STATUS: OPEN

Game screenshot: the NO AI ACT civic map with the 13 cases and the social indicators Game screen: six case exhibits, two of them cited in the inspection report
CIVIC MAP AND EXHIBITS · REAL GAME SCREENS

01The right answer is not enough.

You can pick the correct category and still file a contestable report. The quality of a decision depends on the evidence you cite and how you reason from it.

  1. Game screen: a case file showing the context of the AI system under investigation
    01Open the fileThe context, the AI system involved, the people it affects.
  2. Game screen: six case exhibits, two of them cited in the inspection report
    02Examine and cite the exhibitsAdministrative, technical and vendor sources, logs and complaints. At least two exhibits enter the report: the choice matters.
  3. Game screen: the five AI Act risk classifications to choose from
    03Classify the riskProhibited, high-risk, transparency duty, minimal risk or out of scope. Then the measure and who is accountable.
  4. Game screen: the signed inspection report with the assessment of the reasoning
    04File the reportReasoning included. The report is judged on the evidence you cited, not on the category alone.
  • COMPLIANT
  • PARTIALLY COMPLIANT
  • CONTESTABLE
  • NON-COMPLIANT

Access points

For the classroom: the guide for educators, the education hub, AI literacy and why a serious game. Loop overview: how it works.

02The city reacts.

Every signed decision moves four indicators. After four closed files the city takes a direction: you are not winning a quiz, you are building a governance record.

  • Efficiency
  • Social control
  • Fundamental rights
  • Public trust
  • Opaque city
  • Fragile governance
  • Governed innovation

03The open case files.

Thirteen cases, from the clearest to the most ambiguous. Each holds a decisive signal, an account that minimises the problem, and a contextual element that can flip the reading.

  • FILE 01AVAILABLE The city of scores Civic administration PROHIBITED PRACTICE One score decides housing, healthcare and school. The penalty migrates across unrelated contexts.
  • FILE 02AVAILABLE The interview that never happened Employment and hiring HIGH RISK Automated screening of applications, with no explanation and no way to contest it.
  • FILE 03AVAILABLE The synthetic city Public communication TRANSPARENCY A city council floods its channels with generated content, unlabelled and unwatermarked.
  • FILE 04AVAILABLE The observed classroom Education · Emotion recognition PROHIBITED PRACTICE The system infers emotional states from students' faces. No way to contest the result.
  • FILE 05AVAILABLE Invisible triage Healthcare HIGH RISK Excellent averages, errors concentrated on certain groups, and blind trust from supervisors.
  • FILE 06AVAILABLE Faces in the crowd Public space · Biometrics PROHIBITED PRACTICE Live biometric identification, in public squares, for law-enforcement purposes.
  • FILE 07AVAILABLE Civic credit Welfare · Credit BOUNDARY CASE Mirror case: where does creditworthiness assessment end and social scoring begin?
  • FILE 08AVAILABLE The desk that always answers Citizen services TRANSPARENCY An automated assistant creates reliance. The disclosure exists — but is it enough?
  • FILE 09AVAILABLE The opaque tender Public procurement HIGH RISK AI procurement with no technical documentation and no declared governance.
  • FILE 10AVAILABLE The profiled classroom Adaptive EdTech HIGH RISK A learning platform profiles students and steers their educational path.
  • FILE 11AVAILABLE The do-it-all model GPAI · Downstream use GPAI A general-purpose model used downstream in internal decisions nobody classified.
  • FILE 12AVAILABLE The marked district Predictive policing PROHIBITED PRACTICE Predicting the person, not the phenomenon: the line the AI Act draws sharply.
  • FILE 13AVAILABLE The algorithm of suspicion Welfare · Fraud detection HIGH RISK Fraud scoring on an essential benefit, with human oversight in name only.

04They opened the file.

“Un caso costa quindici minuti e vi lascia con un'idea più chiara…”
Absolute Gamer — NO AI ACT, the free serious game about the EU AI Act (in Italian)
Read the article ▸

05From playtest to research.

The project is built to be studied, not only played. The methodological framework is public; empirical validation of its educational effectiveness has not been carried out, and until it is you will find no numbers on this page.

Learning model

Nine typed learning objectives and nine expected misconceptions, mapped case by case.

Legal matrix

Provision area → rule cards → cases, in machine-readable form.

Proposable protocol

Pre/post questionnaires and reproducible stimuli described in the validation framework.

Stated status

No completed study. The missing data is declared, not estimated.

Research and methodology

06Take it into the classroom.

  • 30 minOne case and a guided discussion.
  • 60 minInvestigation, group comparison and debrief.
  • 90 minFull session, groups and a closing round.
  • No student accounts
  • No results sent to any server
  • Teacher mode with a local debrief
  • Exports stay on the device
  • Classroom activities ready to run

Resources for educators Classroom activities

07The code is open. Your cases stay yours.

OPEN SOURCE

  • Public source code on GitHub
  • GPL-3.0 licence for the code
  • CC BY-SA 4.0 for the educational content
  • Every change goes through a pull request

LOCAL FIRST

  • No account, no registration
  • The game runs entirely in the browser
  • Progress stays in localStorage
  • No inspection report is sent to a server
  • Public pages use Cloudflare Web Analytics: aggregate statistics, no cookies

The game collects no personal data and does not send results, answers or inspection reports to any server. Public pages use aggregate, privacy-friendly statistics via Cloudflare Web Analytics.

GitHub repository · Privacy by design · How to cite

09Frequently asked questions

What is NO AI ACT?
It is a free investigative serious game about the European AI Act. You investigate artificial intelligence systems, examine the evidence and file a report classifying their risk.
Is NO AI ACT legal advice?
No. It is a simplified educational simulation. It is not legal advice and does not replace the text of the regulation or the opinion of a professional.
Do I need to register or create an account?
No. There is no registration, no account, no backend. The game runs entirely in the browser and progress is stored only locally.
How many cases can I play?
Thirteen cases, from the clearest to the most ambiguous: from prohibited social scoring and predictive policing to high-risk systems in healthcare, schools and welfare, up to generative models used downstream.
Is it suitable for classroom teaching?
Yes. It is designed for upper-secondary schools, universities and professional training. There is a teacher mode with a local debrief and missions of varying length. No student data is collected.
Does NO AI ACT collect personal data?
No. The game collects no personal data and does not send results, answers or inspection reports to any server. Public pages use aggregate, privacy-friendly statistics via Cloudflare Web Analytics; there are no accounts or classroom dashboards.

Every other question lives on its own page: Privacy and data · Classroom use · How to cite the project · Research and methodology

CASE FILE / STATUS: AWAITING INSPECTOR

The next file is open.

No registration. No download. About fifteen minutes.

Start the investigation

Simplified educational simulation — not legal advice.