Module 7: Existing and Emerging AI Laws and Standards: Around the Globe Flashcards
Name some jurisdictions that already have laws/guidelines directly related to AI.
- Australia
- Canada
- China
- EU
- South Africa
- South Korea
- UK
What is a key element of Canada’s Artificial Intelligence and Data Act?
- It applies a broad definition of AI to the private sector and identifies many regulated activities that will be affected.
Which jurisdiction has proposed banning AI systems that are excessively risky?
The EU
Which jurisdiction does NOT take a risk based approach to governing AI systems?
China
What is the standard formula for evaluating risk?
Impact x Likelihood
What is a key element of many AI laws to address AI accountability?
Transparency
What are the key element of Canada’s Digital Charter Implementation Act of 2022?
- Adds the proposed Artificial Intelligence and Data Act (AIDA), which builds on existing Canadian consumer protection and human rights law. It is risk-based, focusing on high-impact AI.
- Replaces privacy provisions of PIPEDA with the Consumer Privacy and Protection Act.
- The Personal Information and Data Protection Tribunal Act creates a new data protection tribunal.
What are some of the elements of risks of high impact systems that Canada’s AI laws focus on?
- Nature and severity of the potential harms
- Scale of the use of the system
- Consumer opt-out or control over their interactions with the AI system
- Imbalances of economic and social status of people
- How much autonomy or authority people have to make a different choice
- Degree to which risks are already regulated by industry specific regulations or consumer protection laws.
What apparatus has Canada proposed for AI oversight?
Federal AI and Data Commissioner - Would have the power to:
- Access information
- Order organizations to cease operations (in some circumstances)
Describe Singapore’s approach to AI regulation
Leverage the benefits of AI for productivity and growth while also managing risk and ensuring trust.
What apparatus has Singapore established to govern AI?
An industry led advisory council to advise on the ethical use of AI and data. This council advises the government on issues from commercial uses of AI.
Which country issued Asia’s first AI governance framework? What is the name of the framework and when was it issued?
Singapore issued the Model AI Governance Framework for Traditional AI in 2019.
What is the AI Basic Act?
South Korea’s comprehensive national artificial intelligence regulation, passed in 2025 (second national law after the EU AI Act). It establishes a National Artificial Intelligence Committee to make recommendations to the heads of government agencies, and deliberate and decide on major policies for the development of AI. The act aims to:
- Protect the rights and interests of the people
- Improve the quality of life of the people
- Strengthen national competitiveness
How does the AI Basic Act apply requirements?
The AI Basic Act is more general than the EU AI Act and applies its requirements uniformly across roles and places obligations on “business operators” of the AI, instead of more detailed roles such as providers and deployers.
Business operators are defined as a corporation, organization, individual, or government agency that conducts business related to the AI industry and falls under these categories:
AI Development Business Operator: one who develops and provides AI
AI Utilization Business Operator: one who provides AI products or services
Business operators are subject to:
- Notification requirements
- Risk management and user protection measures
- Explanation and documentation requirements
- Human management requirements
- Impact assessment requirements
Which approach does the AI Basic Act take in regulating AI systems?
The AI Basic Act takes a risk-based approach (similar to the EU AI Act).
“High-impact” AI is defined as an AI system that may have a significant impact on, or pose a risk to, human life, physical safety and basic rights. There are 11 high-risk categories such as health care, management of nuclear materials and production of drinking water. Business operators must review in advance whether the AI is high impact.
What are the obligations of business operators under the AI Basic Act for high impact AI systems?
- Risk management and user protection measures
- Explanation measures for the results derived from AI, the main criteria used to derive the results, and an overview of the learning data used in the development and use of the AI
- Human management and supervision of the AI
- Documentation that can confirm the contents of measures to ensure safety and reliability
- An impact assessment evaluating the impact on the fundamental rights of people
What is the scope and applicability of the AI Basic Act?
Scope: Like the EU AI Act, the AI Basic Act has an extraterritorial scope and applies domestically or internationally if South Korean users or the market are affected.
Applicability: The AI Basic Act excludes artificial intelligence developed and used solely for the purpose of national defense or national security, as prescribed by presidential decree.
Who enforces the AI Basic Act and what are the potential penalties?
Enforcement: The Minister of Science and ICT have the power to conduct investigations and impose and collect fines.
Penalties: The Minister of Science and ICT have the power to conduct investigations and impose and collect fines.
What is the Voluntary Code of Conduct on the Responsible Development and Management of Advanced Generative AI Systems?
Passed in 2023, this is Canada’s common standards for Canadian organizations which enable them to demonstrate that they develop and use generative AI responsibly.
What type of AI regulation does China have in place?
China created AI guidelines in 2023. Key elements:
- Regulatory body: Cyberspace Administration of China
- Applies to services available to the general public in China
- Research institutions are exempt
- Requires generative AI service providers to conduct security reviews and register algorithms with the government if the service can influence public opinion or “mobilize” the public
- Rights-based approach
What are the elements of China’s rights-based approach to AI regulation?
- Overview for end user rights (e.g., clear notice, ability to opt out of personalized recommendations)
- Clear directives for companies using AI for online recommendations, social media and gaming
What are the key elements of Singapore’s Model AI Governance Framework for Traditional AI?
- Voluntary framework based on two principles:
1) decision-making should be explainable, transparent and fair; and
2) AI should be human-centric - AI Verify: toolkit to address AI governance to support testing and oversight
- 2024: Model AI Governance Framework for Generative AI: nine dimensions for creating a trusted environment enabling safe use of generative AI while allowing for innovation
