EU AI Act Risk Assessment Tool

Introduction

This tool will help you determine how your AI system should be classified under the EU AI Act. The assessment will guide you through a series of questions to determine if your system falls into one of these categories:

Unacceptable Risk: AI systems that are prohibited
High Risk: AI systems that require strict compliance with specific requirements
Limited/Transparency Risk: AI systems with transparency obligations
Minimal Risk/General Purpose AI: AI systems with minimal or no specific obligations

Please enter information about your AI solution:

AI Solution Name:

AI Solution Description:

Section 1: Prohibited Practices

First, let's determine if your AI system falls into any prohibited categories under Article 5 of the EU AI Act.

Does your AI system deploy subliminal techniques or use purposefully manipulative or deceptive techniques?

Yes - The system uses subliminal or manipulative techniques No - The system does not use such techniques

Does your AI system exploit vulnerabilities of a specific group (due to age, disability, or social/economic situation)?

Yes - The system exploits such vulnerabilities No - The system does not exploit such vulnerabilities

Does your AI system evaluate or classify natural persons based on social behavior or personal characteristics (social scoring)?

Yes - The system performs social scoring No - The system does not perform social scoring

Does your AI system use biometric identification in publicly accessible spaces for law enforcement purposes?

Yes - The system uses biometric identification in public spaces No - The system does not use biometric identification in public spaces

Does your AI system create or expand facial recognition databases through untargeted scraping?

Yes - The system scrapes facial images without permission No - The system does not scrape facial images

Does your AI system infer emotions in workplace or educational contexts?

Yes - The system infers emotions in these contexts No - The system does not infer emotions in these contexts

Section 2: Biometric Systems

Let's check if your AI system involves biometric identification or categorization as defined in Annex III of the EU AI Act.

Does your AI system perform remote biometric identification?

Yes - The system performs remote biometric identification No - The system does not perform remote biometric identification

Does your AI system perform biometric categorization based on sensitive attributes?

Yes - The system performs biometric categorization No - The system does not perform biometric categorization

Does your AI system perform emotion recognition?

Yes - The system performs emotion recognition No - The system does not perform emotion recognition

Section 3: Critical Infrastructure and Essential Services

Let's determine if your AI system is used in critical infrastructure or essential services as defined in Annex III of the EU AI Act.

Is your AI system used as a safety component in the management or operation of critical infrastructure?

Yes - The system is used in critical infrastructure No - The system is not used in critical infrastructure

Is your AI system used to determine access to essential services?

Yes - The system determines access to essential services No - The system does not determine access to essential services

Is your AI system used to evaluate creditworthiness or establish credit scores?

Yes - The system evaluates creditworthiness No - The system does not evaluate creditworthiness

Is your AI system used for emergency services dispatch or priority decisions?

Yes - The system is used for emergency services dispatch No - The system is not used for emergency services dispatch

Section 4: Education, Employment, and Law Enforcement

Let's check if your AI system is used in education, employment, or law enforcement contexts as defined in Annex III of the EU AI Act.

Is your AI system used in education or vocational training contexts?

Yes - The system is used in education contexts No - The system is not used in education contexts

Is your AI system used in employment, worker management, or recruitment?

Yes - The system is used in employment contexts No - The system is not used in employment contexts

Is your AI system used by law enforcement authorities?

Yes - The system is used by law enforcement No - The system is not used by law enforcement

Is your AI system used in migration, asylum, or border control management?

Yes - The system is used in migration or border control No - The system is not used in migration or border control

Is your AI system used in the administration of justice or democratic processes?

Yes - The system is used in justice or democratic processes No - The system is not used in justice or democratic processes

Section 5: Transparency Requirements

Let's check if your AI system has specific transparency requirements under Article 50 of the EU AI Act.

Does your AI system generate or manipulate text, image, audio, or video content?

Yes - The system generates or manipulates content No - The system does not generate or manipulate content

Does your AI system interact with humans (like chatbots or virtual assistants)?

Yes - The system interacts with humans No - The system does not interact with humans

Does your AI system use emotion recognition or biometric categorization?

Yes - The system uses emotion recognition or biometric categorization No - The system does not use these technologies

How would you best characterize your AI system's purpose and capabilities?

The system performs narrow procedural tasks The system enhances/supports human work without replacing decision-making The system is a general-purpose AI with broad capabilities The system replaces human decision-making in certain contexts The system augments decision-making but humans maintain final control

Section 6: General Purpose AI

Finally, let's determine if your AI system qualifies as General Purpose AI under Article 3 of the EU AI Act.

Is your AI system capable of performing a wide range of distinct tasks?

Yes - The system can perform many distinct tasks No - The system is designed for specific tasks only

How is your AI system deployed or integrated with other systems?

Via API or other programmatic interfaces Through manual integration (not via API) As a standalone system with fixed functionality Embedded within other software or hardware

Does your AI system pose only minimal risks to health, safety, fundamental rights, the environment, democracy, and the rule of law?

Yes - The system poses only minimal risks No - The system may pose more significant risks Uncertain - Risk level depends on specific implementation

Assessment Results

AI System Information

Name:

Description:

Compliance Requirements

Summary of Your Answers

Question	Your Answer	Relevant Article

Save as PDF

You can also save this webpage to your device as a PDF by finding the 'Print' function in your web browser's tools menu.

Press CTRL + P (Windows) or CMD + P (Mac) to open the 'Print' menu.
Select destination: 'Save as PDF'.
Make sure 'Background graphics' (or similar) are enabled in the print settings.
Make sure the headers and footers are unchecked
Click the 'Save' button.