Compliance/Frameworks
EU AI Act Controls: Regulation (EU) 2024/1689
European Union Artificial Intelligence Act | Reference: Regulation (EU) 2024/1689 (phased application 2024-2027) Applicable to: providers, deployers, importers and distributors of AI systems on the EU market
Legend: Critical High Medium Low
Level 0: Prohibited AI (Art. 5): Practices to Avoid
| Code | Title | Priority | Description |
|---|---|---|---|
| EUAI-PROHIB-1 | Prohibition of subliminal manipulation | Critical | Prohibited: AI systems that use subliminal techniques to distort behaviour causing harm |
| EUAI-PROHIB-2 | Prohibition of exploiting vulnerabilities | Critical | Prohibited: systems that exploit the vulnerabilities of specific groups (age, disability, social situation) |
| EUAI-PROHIB-3 | Prohibition of general-purpose social scoring | Critical | Prohibited: social scoring systems by public authorities for general purposes |
| EUAI-PROHIB-4 | Prohibition of real-time biometric identification | Critical | Prohibited (with narrow exceptions): real-time remote biometric identification in publicly accessible spaces |
| EUAI-PROHIB-5 | Prohibition of emotion inference in the workplace | Critical | Prohibited: emotion recognition systems in workplaces and education (with medical exceptions) |
| EUAI-PROHIB-6 | Prohibition of biometric categorisation using sensitive attributes | Critical | Prohibited: categorisation of people based on biometric data by race, political opinions, religion, sexual orientation |
Level 1: High-Risk AI (Art. 6-27)
Provider Obligations (Art. 9-17)
| Code | Title | Priority | Description |
|---|---|---|---|
| EUAI-HR-1 | AI risk management system | Critical | Establish, implement, document and maintain a risk management system for high-risk AI systems |
| EUAI-HR-2 | Data governance and training data | High | Data governance practices for the data sets used in training, validating and testing models |
| EUAI-HR-3 | Technical documentation | High | Draw up complete technical documentation before placing the system on the market (Annex IV) |
| EUAI-HR-4 | Event logging | High | Capability for automatic recording of events during operation for traceability |
| EUAI-HR-5 | Transparency and provision of information | High | Ensure sufficient transparency for deployers to interpret and use the outputs appropriately |
| EUAI-HR-6 | Human oversight | Critical | Design systems to allow effective human oversight during use |
| EUAI-HR-7 | Accuracy, robustness and cybersecurity | Critical | Ensure appropriate levels of accuracy, robustness and resilience against cyberattacks and manipulation |
| EUAI-HR-8 | Quality management system | High | Implement a quality management system for compliance with the AI Act |
| EUAI-HR-9 | Registration in the EU database | High | Register high-risk AI systems in the EU database before placing on the market |
| EUAI-HR-10 | CE marking and declaration of conformity | High | Affix the CE marking and draw up the EU declaration of conformity |
| EUAI-HR-11 | Post-market monitoring | High | Implement and document a post-market monitoring plan for the AI system |
| EUAI-HR-12 | Reporting of serious incidents | Critical | Report serious incidents and malfunctioning to the competent national authorities |
Deployer Obligations (Art. 26)
| Code | Title | Priority | Description |
|---|---|---|---|
| EUAI-OP-1 | Use in accordance with instructions | High | Use high-risk AI systems in accordance with the provider's instructions for use |
| EUAI-OP-2 | Assign a human responsible | Critical | Assign a natural person responsible for the oversight of high-risk AI systems |
| EUAI-OP-3 | Fundamental rights impact assessment | High | Carry out a fundamental rights impact assessment before using high-risk systems |
| EUAI-OP-4 | Keep logs generated by the systems | Medium | Retain the logs automatically generated by high-risk AI systems |
| EUAI-OP-5 | Notify affected persons | High | Inform persons subject to decisions of high-risk AI systems that have a significant impact |
Level 2: General-Purpose AI (GPAI) (Art. 51-56)
| Code | Title | Priority | Description |
|---|---|---|---|
| EUAI-GPAI-1 | Technical documentation of GPAI models | High | Draw up and maintain technical documentation of GPAI models (Annex XI and XII) |
| EUAI-GPAI-2 | Copyright compliance policy | High | Implement a policy to respect copyright in training data |
| EUAI-GPAI-3 | Training data summary | High | Make publicly available a summary of the data used for training |
| EUAI-GPAI-4 | Security policy for models with systemic risk | Critical | For models with potential systemic risk: adversarial evaluation, red team testing, incident management |
| EUAI-GPAI-5 | Incident reporting for models with systemic risk | Critical | Report serious incidents and near misses to the European Commission |
Level 3: Transparency Obligations (Art. 50)
| Code | Title | Priority | Description |
|---|---|---|---|
| EUAI-TRANS-1 | Disclosure of interaction with AI (chatbots) | High | Inform users that they are interacting with an AI system (except where obvious) |
| EUAI-TRANS-2 | Labelling of deepfakes | High | Ensure that synthetic audio, video, image or text content is marked as AI-generated |
| EUAI-TRANS-3 | Disclosure of emotion recognition and biometrics | High | Inform persons exposed to emotion recognition systems or biometric categorisation |
Governance and Oversight
| Code | Title | Priority | Description |
|---|---|---|---|
| EUAI-GOV-1 | Appoint an AI Officer | High | Appoint an internal officer responsible for compliance with the EU AI Act |
| EUAI-GOV-2 | Inventory of AI systems | High | Maintain an inventory of all AI systems used, developed or placed on the market |
| EUAI-GOV-3 | Risk level assessment of AI systems | Critical | Assess and document the risk level of each AI system (prohibited/high/transparency/minimal) |
| EUAI-GOV-4 | EU AI Act training | Medium | Ensure training on the regulation and responsible use of AI for relevant staff |
Application Timetable
| Date | Obligation |
|---|---|
| Feb 2025 | Prohibited AI practices (Art. 5): already applicable |
| Aug 2025 | GPAI and general-purpose models (Art. 51-56) |
| Aug 2026 | High-risk AI systems (majority) + transparency obligations |
| Aug 2027 | High-risk AI systems in pre-existing regulated infrastructure |
Competent authority in Portugal: To be designated (presumably the CNPD or a new authority) | Coordination: European Commission / AI Office
Total: 38 controls in the catalogue
Source: Regulation (EU) 2024/1689 (AI Act), published in the OJEU on 12 July 2024