ai_system_name stringlengths 14 47 | company_example stringlengths 6 48 | country_of_origin stringlengths 2 15 | risk_tier stringclasses 5
values | article_reference stringlengths 10 34 ⌀ | domain stringlengths 5 23 | use_case stringlengths 47 83 | compliance_deadline stringdate 2025-02-02 00:00:00 2027-08-02 00:00:00 ⌀ | compliance_status stringclasses 7
values | fine_tier_eur stringclasses 5
values | requires_conformity_assessment bool 2
classes | requires_eu_database_registration bool 2
classes | human_oversight_required bool 2
classes | transparency_obligation bool 2
classes | sector stringlengths 5 30 | affected_fundamental_rights stringlengths 16 61 | deployment_countries_eu int64 3 27 | estimated_eu_users_affected int64 0 480M | compliance_cost_estimate_eur int64 0 85M | market_withdrawal_required bool 2
classes | gpai_model bool 2
classes | systemic_risk bool 2
classes | enforcement_active bool 2
classes | key_risk stringlengths 7 49 | annual_revenue_usd_bn float64 0 514 | days_to_deadline int64 -400 511 | deadline_days int64 -400 511 | compliance_urgency stringclasses 3
values |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Social Credit Scoring System | ByteDance (Sesame Credit equivalent) | China | Prohibited | Article 5(1)(c) | Social Scoring | Scoring individuals based on social behavior and trustworthiness | 2025-02-02 | Banned | 35M or 7% turnover | false | false | false | false | Government/Finance | Human dignity, Non-discrimination, Privacy | 3 | 0 | 0 | true | false | true | true | Discriminatory social control | 110 | -400 | -400 | OVERDUE |
Subliminal Manipulation Engine | Cambridge Analytica-style systems | USA | Prohibited | Article 5(1)(a) | Behavioral Manipulation | Using subliminal techniques to distort human behavior against their will | 2025-02-02 | Banned | 35M or 7% turnover | false | false | false | false | Political/Advertising | Autonomy, Free will, Democracy | 12 | 0 | 0 | true | false | true | true | Manipulation of democratic processes | 0 | -400 | -400 | OVERDUE |
Real-Time Public Facial Recognition | Clearview AI | USA | Prohibited | Article 5(1)(h) | Biometrics | Real-time remote biometric identification in public spaces by private entities | 2025-02-02 | Banned | 35M or 7% turnover | false | false | false | false | Law Enforcement/Security | Privacy, Non-discrimination, Free movement | 18 | 0 | 0 | true | false | true | true | Mass surveillance | 0.03 | -400 | -400 | OVERDUE |
Vulnerability Exploitation AI | Dark pattern ad targeting systems | Various | Prohibited | Article 5(1)(b) | Behavioral Manipulation | Exploiting age, disability, or socioeconomic vulnerability to distort behavior | 2025-02-02 | Banned | 35M or 7% turnover | false | false | false | false | Advertising/Retail | Human dignity, Protection of vulnerable groups | 27 | 0 | 0 | true | false | true | true | Exploitation of vulnerable persons | 0 | -400 | -400 | OVERDUE |
Workplace Emotion Recognition System | Affectiva (iMotions) | USA | Prohibited | Article 5(1)(f) | Biometrics | Inferring emotions of employees in workplace settings | 2025-02-02 | Banned | 35M or 7% turnover | false | false | false | false | HR/Enterprise | Privacy, Worker rights, Dignity | 15 | 0 | 0 | true | false | false | true | Invasive workplace surveillance | 0.1 | -400 | -400 | OVERDUE |
Classroom Emotion Surveillance | HonorLock, Proctorio (emotion module) | USA | Prohibited | Article 5(1)(f) | Education | Emotion recognition systems deployed in educational institutions | 2025-02-02 | Banned | 35M or 7% turnover | false | false | false | false | Education | Privacy, Children rights, Dignity | 8 | 0 | 0 | true | false | false | true | Mass student surveillance | 0.05 | -400 | -400 | OVERDUE |
Biometric Characteristic Inference AI | Faception, PimEyes | Israel/Poland | Prohibited | Article 5(1)(g) | Biometrics | Inferring race, political opinion, religion, sexual orientation from biometrics | 2025-02-02 | Banned | 35M or 7% turnover | false | false | false | false | Security/Law Enforcement | Non-discrimination, Privacy, Human dignity | 6 | 0 | 0 | true | false | true | true | Discriminatory profiling | 0.02 | -400 | -400 | OVERDUE |
Untargeted Facial Recognition Scraping | Clearview AI (scraping module) | USA | Prohibited | Article 5(1)(e) | Biometrics | Mass scraping of internet/CCTV images to build facial recognition databases | 2025-02-02 | Banned | 35M or 7% turnover | false | false | false | false | Law Enforcement | Privacy, Data protection, Freedom of assembly | 5 | 0 | 0 | true | false | true | true | Covert mass biometric database creation | 0.03 | -400 | -400 | OVERDUE |
Biometric Verification System (Airport) | IDEMIA | France | High Risk | Annex III, Item 1 | Biometrics | Biometric verification of travellers at border control | 2026-08-02 | Must Comply | 15M or 3% turnover | true | true | true | true | Travel/Security | Privacy, Free movement | 22 | 45,000,000 | 4,200,000 | false | false | false | false | Biometric data misuse, false positives | 1.8 | 146 | 146 | CRITICAL (<6mo) |
Live Biometric Identification (Law Enforcement) | Thales Group | France | High Risk | Annex III, Item 1a | Biometrics | Post-facto biometric identification for law enforcement investigations | 2026-08-02 | Must Comply | 15M or 3% turnover | true | true | true | true | Law Enforcement | Privacy, Presumption of innocence | 14 | 8,000,000 | 7,500,000 | false | false | false | false | Wrongful identification, civil liberties | 19.4 | 146 | 146 | CRITICAL (<6mo) |
Smart Grid AI Management System | Siemens Energy AI | Germany | High Risk | Annex III, Item 2 | Critical Infrastructure | AI management of electricity grid load balancing and failure prevention | 2026-08-02 | In Progress | 15M or 3% turnover | true | true | true | true | Energy | Safety, Access to essential services | 19 | 80,000,000 | 9,800,000 | false | false | true | false | Grid failure, cascading infrastructure collapse | 36 | 146 | 146 | CRITICAL (<6mo) |
Water Treatment AI Monitor | Veolia AI Systems | France | High Risk | Annex III, Item 2 | Critical Infrastructure | Automated monitoring and control of municipal water treatment plants | 2026-08-02 | In Progress | 15M or 3% turnover | true | true | true | true | Water/Utilities | Right to life, Health, Safety | 11 | 35,000,000 | 5,600,000 | false | false | true | false | Contamination risk if AI fails | 7.2 | 146 | 146 | CRITICAL (<6mo) |
AI Admission Scoring System | Uniaid, various university tools | UK | High Risk | Annex III, Item 3 | Education | Automated scoring and ranking of university applicants | 2026-08-02 | Non-Compliant | 15M or 3% turnover | true | true | true | true | Education | Non-discrimination, Right to education, Fairness | 16 | 2,800,000 | 1,200,000 | false | false | false | false | Algorithmic bias in university access | 0.15 | 146 | 146 | CRITICAL (<6mo) |
AI Student Performance Prediction | Knewton (Wiley) | USA | High Risk | Annex III, Item 3 | Education | Predicting student academic outcomes to determine course placement | 2026-08-02 | In Progress | 15M or 3% turnover | true | true | true | true | Education | Non-discrimination, Education rights | 20 | 5,500,000 | 2,100,000 | false | false | false | false | Reinforcing educational inequality | 1.9 | 146 | 146 | CRITICAL (<6mo) |
AI CV Screening & Recruitment Tool | HireVue, Workday AI | USA | High Risk | Annex III, Item 4 | Employment | Automated filtering and ranking of job applicants from CVs | 2026-08-02 | Non-Compliant | 15M or 3% turnover | true | true | true | true | HR/Recruitment | Non-discrimination, Right to work, Fairness | 27 | 28,000,000 | 3,400,000 | false | false | false | false | Gender/race discrimination in hiring | 0.45 | 146 | 146 | CRITICAL (<6mo) |
AI Employee Monitoring & Productivity Scoring | Microsoft Viva Insights, Aware | USA | High Risk | Annex III, Item 4b | Employment | Monitoring worker behavior and scoring productivity for management decisions | 2026-08-02 | In Progress | 15M or 3% turnover | true | true | true | true | Enterprise Software | Privacy, Worker dignity, Fair treatment | 27 | 65,000,000 | 8,900,000 | false | false | false | false | Surveillance chilling effect, burnout | 211.9 | 146 | 146 | CRITICAL (<6mo) |
AI Gig Worker Task Allocation System | Uber Dispatch AI, Deliveroo Algorithm | USA/UK | High Risk | Annex III, Item 4a | Employment | Automated allocation of work tasks and management of gig economy workers | 2026-08-02 | Non-Compliant | 15M or 3% turnover | true | true | true | true | Gig Economy/Transport | Fair working conditions, Non-discrimination | 23 | 8,500,000 | 5,700,000 | false | false | false | false | Opaque deactivation, unfair task distribution | 37.3 | 146 | 146 | CRITICAL (<6mo) |
AI Credit Scoring System | FICO, Experian AI, Schufa | USA/Germany | High Risk | Annex III, Item 5b | Financial Services | Automated creditworthiness assessment for loans and credit products | 2026-08-02 | In Progress | 15M or 3% turnover | true | true | true | true | Finance/Banking | Non-discrimination, Right to explanation, Financial inclusion | 27 | 120,000,000 | 15,000,000 | false | false | true | false | Discriminatory lending, financial exclusion | 1.4 | 146 | 146 | CRITICAL (<6mo) |
AI Insurance Risk Assessment | Lemonade AI, Zurich Insurance AI | USA/Switzerland | High Risk | Annex III, Item 5b | Insurance | Automated risk scoring for insurance pricing and coverage decisions | 2026-08-02 | In Progress | 15M or 3% turnover | true | true | true | true | Insurance | Non-discrimination, Access to services | 22 | 95,000,000 | 6,800,000 | false | false | false | false | Discriminatory pricing by demographics | 2.1 | 146 | 146 | CRITICAL (<6mo) |
Social Benefit Eligibility AI | Various EU government systems | EU (various) | High Risk | Annex III, Item 5a | Public Services | AI determining eligibility for social welfare, housing benefits, unemployment aid | 2026-08-02 | Non-Compliant | 15M or 3% turnover | true | true | true | true | Government | Social rights, Non-discrimination, Dignity | 21 | 42,000,000 | 11,000,000 | false | false | true | false | Wrongful benefit denial to vulnerable groups | 0 | 146 | 146 | CRITICAL (<6mo) |
Predictive Policing System | PredPol (Geolitica), SAS Crime AI | USA | High Risk | Annex III, Item 6 | Law Enforcement | AI predicting crime hotspots and individual crime likelihood for policing decisions | 2026-08-02 | Non-Compliant | 15M or 3% turnover | true | true | true | true | Law Enforcement | Presumption of innocence, Non-discrimination, Fair trial | 9 | 18,000,000 | 4,400,000 | false | false | true | false | Racial profiling, self-fulfilling crime loops | 0.08 | 146 | 146 | CRITICAL (<6mo) |
AI Lie Detection / Credibility Assessment | iBorderCtrl (EU-funded, discontinued) | EU | High Risk | Annex III, Item 6a | Law Enforcement | AI assessment of truthfulness or deception in interrogations or border checks | 2026-08-02 | Non-Compliant | 15M or 3% turnover | true | true | true | true | Border Control/Law Enforcement | Fair trial, Presumption of innocence, Human dignity | 4 | 500,000 | 900,000 | false | false | false | false | Pseudoscience-based wrongful accusation | 0.01 | 146 | 146 | CRITICAL (<6mo) |
AI Asylum Application Risk Classifier | EU Member State border AI tools | EU (various) | High Risk | Annex III, Item 7 | Migration | Automated risk assessment and classification of asylum seekers | 2026-08-02 | Non-Compliant | 15M or 3% turnover | true | true | true | true | Border Control | Right to asylum, Non-refoulement, Non-discrimination | 13 | 1,200,000 | 3,800,000 | false | false | false | false | Wrongful deportation, rights violations | 0 | 146 | 146 | CRITICAL (<6mo) |
AI Legal Research & Case Outcome Predictor | Kira Systems, Luminance, Lex Machina | Canada/UK/USA | High Risk | Annex III, Item 8 | Justice | AI predicting judicial outcomes and assisting in legal interpretation | 2026-08-02 | In Progress | 15M or 3% turnover | true | true | true | true | Legal | Fair trial, Rule of law, Access to justice | 18 | 3,500,000 | 2,900,000 | false | false | false | false | Algorithmic influence on judicial independence | 0.28 | 146 | 146 | CRITICAL (<6mo) |
AI Diagnostic Imaging System (Radiology) | Siemens Healthineers AI-Rad, Aidoc | Germany/Israel | High Risk | Annex I (MDR) + Article 6(1) | Healthcare | AI-assisted detection of tumours, fractures, and pathologies in medical imaging | 2027-08-02 | In Progress | 15M or 3% turnover | true | true | true | true | Healthcare | Right to life, Health, Informed consent | 24 | 22,000,000 | 18,000,000 | false | false | false | false | Misdiagnosis, false negatives in cancer detection | 21 | 511 | 511 | FUTURE |
AI Clinical Decision Support System | IBM Watson Health (DoQuMed), Philips IntelliSite | USA/Netherlands | High Risk | Annex I (MDR) + Article 6(1) | Healthcare | AI recommending drug dosages and treatment protocols to clinicians | 2027-08-02 | In Progress | 15M or 3% turnover | true | true | true | true | Healthcare | Right to life, Health, Informed consent | 21 | 31,000,000 | 24,000,000 | false | false | false | false | Incorrect treatment recommendation | 7.8 | 511 | 511 | FUTURE |
AI Mental Health Triage System | Wysa, Limbic (NHS-approved) | India/UK | High Risk | Annex III, Item 5 + MDR | Healthcare | AI assessing mental health severity and routing patients to appropriate care | 2026-08-02 | In Progress | 15M or 3% turnover | true | true | true | true | Healthcare | Right to health, Dignity, Privacy | 12 | 9,500,000 | 3,100,000 | false | true | false | false | Missed suicidal ideation, data misuse | 0.08 | 146 | 146 | CRITICAL (<6mo) |
Autonomous Vehicle Level 3+ AI | Waymo, Mobileye (Intel) | USA/Israel | High Risk | Annex I (Vehicle Type Approval) | Automotive | Autonomous driving and collision avoidance in road vehicles | 2027-08-02 | In Progress | 15M or 3% turnover | true | true | true | true | Automotive | Right to life, Safety | 10 | 4,200,000 | 85,000,000 | false | false | false | false | Fatal accident from AI failure | 4 | 511 | 511 | FUTURE |
ChatGPT / Claude / Gemini (Consumer) | OpenAI, Anthropic, Google | USA | Limited Risk | Article 50 | General Purpose AI | General-purpose AI chatbot for consumer and enterprise use | 2026-08-02 | In Progress | 7.5M or 1% turnover (misinformation) | false | false | false | true | Technology | Transparency, Informed consent | 27 | 185,000,000 | 2,800,000 | false | true | true | false | Misinformation, lack of AI disclosure | 3.7 | 146 | 146 | CRITICAL (<6mo) |
AI Customer Service Chatbot | Intercom Fin, Zendesk AI | USA | Limited Risk | Article 50(1) | Customer Service | AI chatbot interacting with customers for support and sales | 2026-08-02 | Compliant | 7.5M or 1% turnover | false | false | false | true | Enterprise Software | Transparency, Consumer rights | 27 | 210,000,000 | 450,000 | false | true | false | false | Consumers unaware they interact with AI | 1.9 | 146 | 146 | CRITICAL (<6mo) |
AI-Generated News Content Tool | Associated Press Automated Insights | USA | Limited Risk | Article 50(4) | Media | AI generating news articles, summaries and editorial content | 2026-08-02 | Compliant | 7.5M or 1% turnover | false | false | false | true | Media | Freedom of information, Democratic debate | 24 | 95,000,000 | 320,000 | false | true | false | false | AI-generated misinformation at scale | 0.5 | 146 | 146 | CRITICAL (<6mo) |
Deepfake Detection & Generation Platform | Synthesia, D-ID, ElevenLabs | UK/Israel/USA | Limited Risk | Article 50(2)(4) | Media/Content | AI video and voice synthesis (deepfake) for content creation | 2026-08-02 | In Progress | 7.5M or 1% turnover | false | false | false | true | Media/Entertainment | Right to truth, Reputation, Non-manipulation | 27 | 62,000,000 | 890,000 | false | true | false | false | Non-consensual deepfakes, election interference | 0.5 | 146 | 146 | CRITICAL (<6mo) |
AI Recommendation Algorithm (Social Media) | TikTok FYP, Instagram Reels AI | China/USA | Limited Risk | Article 50 + DSA | Social Media | AI content recommendation and feed curation for social media users | 2026-08-02 | In Progress | 7.5M or 1% turnover | false | false | false | true | Social Media | Autonomy, Mental health, Privacy | 27 | 260,000,000 | 4,200,000 | false | true | true | false | Addictive design, radicalization pipeline | 16.4 | 146 | 146 | CRITICAL (<6mo) |
AI Spam Filter | Google Gmail AI, Microsoft Defender | USA | Minimal Risk | null | Email/Security | Automated filtering of spam and phishing emails | null | Exempt | null | false | false | false | false | Technology | None significant | 27 | 320,000,000 | 0 | false | false | false | false | Minimal | 280 | 0 | 0 | CRITICAL (<6mo) |
AI-Enabled Video Game NPC | Inworld AI (Nvidia ACE integration) | USA | Minimal Risk | null | Gaming | AI-driven non-player characters with adaptive behavior in video games | null | Exempt | null | false | false | false | false | Gaming | None significant | 27 | 89,000,000 | 0 | false | true | false | false | Minimal | 0.15 | 0 | 0 | CRITICAL (<6mo) |
AI Product Recommendation Engine | Amazon Personalize, Netflix AI | USA | Minimal Risk | null | E-commerce | Personalized product and content recommendations for online shoppers | null | Exempt | null | false | false | false | false | E-commerce/Media | None significant | 27 | 290,000,000 | 0 | false | false | false | false | Minimal — filter bubbles | 514 | 0 | 0 | CRITICAL (<6mo) |
AI Translation Tool | DeepL, Google Translate | Germany/USA | Minimal Risk | null | Language | Automated language translation for consumers and businesses | null | Exempt | null | false | false | false | false | Technology | None significant | 27 | 480,000,000 | 0 | false | true | false | false | Minimal — mistranslation risk | 0.08 | 0 | 0 | CRITICAL (<6mo) |
AI Code Assistant | GitHub Copilot, Cursor, Tabnine | USA | Minimal Risk | null | Developer Tools | AI-assisted code generation and completion for software developers | null | Exempt | null | false | false | false | false | Technology | None significant | 27 | 45,000,000 | 0 | false | true | false | false | Copyright/IP concerns | 2 | 0 | 0 | CRITICAL (<6mo) |
GPT-4 / GPT-4o (API) | OpenAI | USA | GPAI — Systemic Risk | Article 51, Chapter V | General Purpose AI | Foundation model API for enterprise AI applications | 2025-08-02 | In Progress | 15M or 3% turnover | true | true | false | true | Technology | Privacy, Autonomy, Democracy | 27 | 95,000,000 | 42,000,000 | false | true | true | true | Systemic misuse, mass manipulation | 3.7 | -219 | -219 | OVERDUE |
Gemini Ultra / Gemini 1.5 Pro | Google DeepMind | USA | GPAI — Systemic Risk | Article 51, Chapter V | General Purpose AI | Multimodal foundation model for search, workspace and developer APIs | 2025-08-02 | In Progress | 15M or 3% turnover | true | true | false | true | Technology | Privacy, Autonomy, Democracy | 27 | 180,000,000 | 38,000,000 | false | true | true | true | Systemic bias, search monopoly | 307 | -219 | -219 | OVERDUE |
Claude 3.5 / Claude Sonnet 4 | Anthropic | USA | GPAI — Systemic Risk | Article 51, Chapter V | General Purpose AI | Foundation AI model for enterprise and consumer applications | 2025-08-02 | Compliant | 15M or 3% turnover | true | true | false | true | Technology | Privacy, Autonomy | 27 | 42,000,000 | 28,000,000 | false | true | true | true | Systemic misuse potential | 1 | -219 | -219 | OVERDUE |
Llama 3 / Llama 4 (Open Source) | Meta AI | USA | GPAI — Systemic Risk | Article 53 (Open Source Exemption) | General Purpose AI | Open-source foundation model for unrestricted deployment | 2025-08-02 | Partial Exemption | Reduced — open source exemption | false | true | false | true | Technology | Privacy, Safety, Autonomy | 27 | 55,000,000 | 8,500,000 | false | true | true | true | Uncontrolled deployment by third parties | 134.9 | -219 | -219 | OVERDUE |
EU AI Act Risk Register 2026 | 42 AI Systems
42 AI systems classified by risk tier, article & deadline
Dataset Info
- Source: Kaggle
- Original Size: 0.01 MB
- Kaggle Downloads: 59
- Files: 1
Files
eu_ai_act_risk_register_2026.csv
Mirrored from Kaggle
- Downloads last month
- 22