Regulation Types
AI governance takes many forms - from binding laws to voluntary guidelines. Understanding these types helps you assess the legal weight of different instruments.
Quick Reference
Legally Binding
Must be followed; violations have legal consequences
Non-Binding
Advisory; may influence future binding rules
Proposals
Not yet law; may be enacted or rejected
Act
(Primary Law)Legislation passed by a legislature (e.g., Acts of Parliament, Codes, Statutory Laws, Organic Laws).
Key Characteristics:
- Passed through full legislative process
- Highest level of legal authority
- Requires parliamentary/congressional approval
- Often sets broad legal frameworks
Examples:
EU AI Act, UK Online Safety Act, California Consumer Privacy Act, German AI Act
Bill
(Proposal)Draft legislation formally introduced to a legislature but not yet enacted.
Key Characteristics:
- Formally registered with legislative body
- Subject to debate and amendment
- May pass or fail to become law
- Indicates legislative intent
Examples:
US AI Bill of Rights (proposed), Canada AIDA Bill C-27, Brazil AI Bill
Decree
(Ordinance)Law issued by the Executive/Government with the force of law, often bypassing immediate parliamentary debate.
Key Characteristics:
- Issued by executive authority
- Immediate legal effect
- May require later parliamentary ratification
- Common in civil law jurisdictions
Examples:
French AI Decrees, Italian Emergency Decrees (OUG), US Executive Orders on AI
Regulation
(Binding Rules)Binding legal rules. Covers both Agency Rules (issued by regulators like SEC/EPA) and Supra-national Laws (like EU Regulations).
Key Characteristics:
- Directly applicable and binding
- Issued by regulatory agencies or supranational bodies
- Often implements broader legislation
- Detailed technical requirements
Examples:
EU AI Act (as Regulation), FTC AI Rules, SEC AI Disclosure Rules
Policy
(Strategy)High-level government documents outlining intent, national strategy, or action plans (not legally binding).
Key Characteristics:
- Sets government direction and priorities
- Not legally enforceable
- Guides future legislation and investment
- Often includes timelines and targets
Examples:
National AI Strategies, AI Action Plans, Digital Economy Policies
Guideline
(Framework)Non-binding recommendations, best practices, or structural principles issued by authorities.
Key Characteristics:
- Advisory in nature
- May become de facto standards
- Often precedes binding regulation
- Provides compliance guidance
Examples:
NIST AI Risk Management Framework, OECD AI Principles, Singapore Model AI Governance Framework
Standard
(Technical Specification)Technical specifications or metrics (e.g., ISO, IEC, NIST standards).
Key Characteristics:
- Precise technical requirements
- Often developed by standards bodies
- May be referenced in binding regulation
- Industry-driven development
Examples:
ISO/IEC 42001 (AI Management), IEEE AI Standards, CEN/CENELEC AI Standards