Recitals

• Recital 1: Purpose of the regulation
• Recital 2: Compatibility with the values of the European Union
• Recital 3: Establishing a uniform level of protection for AI in the EU
• Recital 4: Definition of artificial intelligence
• Recital 5: Possible risks associated with the use of AI
• Recital 6: People-centered development in line with EU values
• Recital 7: Level of protection
• Recital 8: Legal framework for AI in the EU internal market
• Recital 9: Harmonized rules for high-risk AI systems in the EU legal framework
• Recital 10: Validity of existing legal bases
• Recital 11: Liability of intermediaries under Directive 2000/31/EC
• Recital 12: Definition and characteristics of AI systems
• Recital 13: Operator
• Recital 14: Biometric data
• Recital 15: Biometric identification
• Recital 16: Biometric categorization
• Recital 17: Biometric remote identification system
• Recital 18: Emotion recognition system
• Recital 19: Publicly accessible space
• Recital 20: Necessary concepts for awareness-raising and AI competence in the EU
• Recital 21: Equal opportunity application of the regulation for AI providers
• Recital 22: Applicability for actors in third countries
• Recital 23: Applicability to EU institutions, authorities and other bodies
• Recital 24: Exceptional areas military, defense and national security
• Recital 25: Exceptional areas research and development
• Recital 26: Risk-based approach
• Recital 27: Principles for AI development
• Recital 28: Harmful or improper use
• Recital 29: Influencing human behavior
• Recital 30: Use of AI for biometric categorization
• Recital 31: Use of AI for social assessment
• Recital 32: Use of AI for real-time remote identification
• Recital 33: Use of biometric real-time remote identification systems for law enforcement purposes
• Recital 34: Proportionate use and fundamental rights impact assessment
• Recital 35: Authorization requirement
• Recital 36: Information of responsible authorities
• Recital 37: Adoption into national law
• Recital 38: Processing of biometric data
• Recital 39: Applicability of Directive (EU) 2016/680
• Recital 40: Binding effect for Ireland
• Recital 41: Binding effect for Denmark
• Recital 42: Presumption of innocence
• Recital 43: Ban on mass surveillance using AI facial recognition
• Recital 44: Risks and limits of emotion recognition
• Recital 45: Further Union law remains unaffected
• Recital 46: Safety requirements for high-risk systems
• Recital 47: Risk minimization
• Recital 48: Protection of fundamental rights
• Recital 49: Adaptation of legislation
• Recital 50: Classification as a high-risk system
• Recital 51: Effects of classification as a high-risk system
• Recital 52: Criteria for classification as a high-risk system
• Recital 53: Exceptions and risks for the risk classification
• Recital 54: Use cases and risk assessment
• Recital 55: Management and operation of critical infrastructure
• Recital 56: Use in the area of education
• Recital 57: Use in the area of employment, personnel management and self-employment
• Recital 58: Use for social benefits, creditworthiness and emergencies
• Recital 59: Use in law enforcement
• Recital 60: Use in the area of migration, asylum and border control
• Recital 61: Use in the administration of justice
• Recital 62: Use in elections
• Recital 63: Legality under other EU legal acts
• Recital 64: Compatibility with other Union law
• Recital 65: Use of risk management system
• Recital 66: Scope of the requirements
• Recital 67: Data governance and data management
• Recital 68: Access to high-quality data
• Recital 69: Right to data protection and privacy
• Recital 70: Right to protection from discrimination
• Recital 71: Development documentation
• Recital 72: Transparency
• Recital 73: Monitoring by natural persons
• Recital 74: Key performance indicator level
• Recital 75: Technical robustness
• Recital 76: Cybersecurity
• Recital 77: Safety requirements according to the regulation
• Recital 78: Conformity assessment procedure
• Recital 79: Responsibility
• Recital 80: Equality and special protection
• Recital 81: Quality management system
• Recital 82: Level playing field
• Recital 83: Roles and duties of the players
• Recital 84: Validity as a provider of systems under certain conditions
• Recital 85: Interaction and cooperation between providers of general-purpose high-risk AI systems
• Recital 86: Cooperation of former providers
• Recital 87: AI system as a safety component
• Recital 88: Provision of information
• Recital 89: Release from responsibilities
• Recital 90: Model contract terms
• Recital 91: Responsibilities and obligations for operators
• Recital 92: Duty to inform and consult employees and representatives
• Recital 93: Informing affected parties
• Recital 94: Application of Directive (EU) 2016/680 for the processing of biometric data for law enforcement purposes
• Recital 95: Protection regulations for subsequent remote biometric identification
• Recital 96: Fundamental rights impact assessment
• Recital 97: Definition of individual AI models
• Recital 98: Model determination
• Recital 99: Generative AI models as a general purpose model
• Recital 100: Examples of AI model with general purpose
• Recital 101: Transparency obligations for providers
• Recital 102: Transparency obligations for open source licenses
• Recital 103: Definition of an AI model with general purpose and open source license
• Recital 104: Exemptions from the transparency obligation
• Recital 105: Use of protected content during development
• Recital 106: Warranty for the fulfillment of obligations by providers
• Recital 107: Transparency obligations for training data
• Recital 108: Compliance control
• Recital 109: Proportionality of surveillance measures
• Recital 110: Systemic risks for general-purpose AI models
• Recital 111: Classification of AI models with systemic risk
• Recital 112: Clarification of the procedure for classification
• Recital 113: Power of the Commission to classify
• Recital 114: Risk assessment obligation for providers
• Recital 115: Risk mitigation obligation
• Recital 116: Codes of conduct for AI models with general purpose and systemic risks
• Recital 117: Uniform codes of conduct by the EU Commission
• Recital 118: Regulation of AI systems in the Union by regulation
• Recital 119: AI models as intermediary services
• Recital 120: Importance of obligations for providers and operators
• Recital 121: Harmonized standards
• Recital 122: Compliance with cyber security requirements
• Recital 123: Conformity assessment of high-risk systems
• Recital 124: Assessment of compliance with the requirements of the regulation
• Recital 125: Limited conformity assessment by third parties
• Recital 126: Procedure for conformity assessment by third parties
• Recital 127: Recognition of conformity assessment results
• Recital 128: Renewed conformity assessment in the event of significant changes
• Recital 129: CE marking for high-risk systems
• Recital 130: Specific grounds for placing on the market without conformity assessment
• Recital 131: Obligation to register high-risk models
• Recital 132: Transparency obligations where there is a particular risk of fraud and identity theft
• Recital 133: Mandatory labeling of AI-generated results
• Recital 134: Obligation to disclose AI-generated content
• Recital 135: Codes of conduct for AI content review
• Recital 136: Importance of transparency for the implementation of the regulation
• Recital 137: Transparency and legality
• Recital 138: Introduction of real laboratories
• Recital 139: Goals of real laboratories
• Recital 140: Legal basis for the use of personal data in the public interest
• Recital 141: Testing AI systems under real conditions
• Recital 142: Encouraging support and promotion by member states
• Recital 143: Consideration of SMEs, including start-ups
• Recital 144: Promoting and protecting innovation
• Recital 145: Minimizing risk and facilitating compliance with obligations
• Recital 146: Relief for micro-enterprises
• Recital 147: Facilitating access for test and experimental facilities
• Recital 148: Introduction of a governance framework
• Recital 149: Establishment of a committee
• Recital 150: Advisory forum
• Recital 151: Scientific committee
• Recital 152: Establishment of Union structures to support AI systems
• Recital 153: Designation of at least one notifying authority and one monitoring authority per member state
• Recital 154: Powers of competent national authorities
• Recital 155: System for monitoring high-risk models and reporting serious incidents
• Recital 156: Validity of the system for market surveillance and conformity of products
• Recital 157: Position of national authorities or other bodies
• Recital 158: Supervision and market surveillance of AI systems in the financial sector
• Recital 159: Powers of competent authorities for biometric data
• Recital 160: Joint activities of market surveillance authorities
• Recital 161: Responsibilities and competencies
• Recital 162: Synergies at Union level
• Recital 163: Scientific panel as supporter of the Office for AI
• Recital 164: Competencies of the AI Office
• Recital 165: Voluntary codes of conduct for non-high-risk AI systems
• Recital 166: Safety of non-high-risk AI systems
• Recital 167: Cooperation between competent authorities
• Recital 168: Implementation of provisions
• Recital 169: Fines
• Recital 170: Legal remedies
• Recital 171: Declaration on the decision of high-risk systems
• Recital 172: Whistleblower protection
• Recital 173: Powers of the Commission after delegation
• Recital 174: Assessment and review
• Recital 175: Delegation of executive powers
• Recital 176: Principle of proportionality and subsidiarity
• Recital 177: Ensuring continuity
• Recital 178: Voluntary compliance during the transition phase
• Recital 179: Date of validity
• Recital 180: Decree