The global AI law landscape
The exam tests recognition of the major patterns more than every detail. Most laws repeat a few ideas: risk tiers, impact assessments, transparency notices, bias testing and human review.
United States: federal
- No comprehensive federal AI statute.
- EO 14110 (Oct 2023, on safe, secure and trustworthy AI) was revoked in January 2025. Federal policy now emphasizes innovation and competitiveness (America's AI Action Plan, July 2025).
- Sector regulators still enforce existing law: FTC, EEOC, CFPB, FDA (AI-enabled medical devices), SEC, HUD.
United States: states and cities
| Measure | Key points |
|---|---|
| Colorado (SB 26-189, replacing SB 24-205) | The 2024 AI Act never took effect: delayed to 30 June 2026, then repealed and replaced on 14 May 2026. The new law covers automated decision-making technology that materially influences a consequential decision (education, employment, housing, lending, insurance, health care, essential government services). From 1 January 2027 developers give deployers documentation; deployers give notice at the point of interaction and, after an adverse outcome, a plain-language explanation within 30 days; consumers can see and correct the personal data used and ask for meaningful human review. The attorney general enforces it; there is no private right of action. Know the 2024 version too (reasonable care against algorithmic discrimination, impact assessments, an affirmative defense for NIST or ISO alignment), since older questions may test it |
| NYC Local Law 144 | Automated employment decision tools need an annual independent bias audit, a public summary and candidate notice |
| Illinois | BIPA (biometric consent, private right of action); AI Video Interview Act; 2024 amendment to the Human Rights Act on AI in employment |
| California | CCPA automated decision-making technology regulations; SB 53 frontier AI transparency (2025); laws on AI-generated content and deepfakes |
| Utah AI Policy Act | Disclosure when consumers interact with generative AI |
| Texas TRAIGA | Effective 1 January 2026; government AI duties and prohibited uses |
South Korea: the AI Basic Act
The Body of Knowledge names it alongside the EU AI Act. In force since 22 January 2026, it's the second comprehensive AI law after the EU's, and it reaches foreign providers whose AI affects Korean users or markets.
| Obligation | Who and what |
|---|---|
| Advance notice | Tell users beforehand that a product or service uses high-impact or generative AI |
| Generative AI labeling | Label AI-generated output; give a clear notice for synthetic audio, images or video that are hard to tell from real ones |
| High-impact AI | AI in sensitive areas such as health care, energy, transport, hiring and biometric analysis: explain outcomes and their main criteria, run a user protection plan, keep human oversight, keep records, and try to assess the impact on fundamental rights |
| High-performance AI | Models trained with at least 10^26 operations: manage risk across the lifecycle and report to the ministry |
| Domestic representative | Foreign companies above revenue or user thresholds must appoint one in Korea |
The Ministry of Science and ICT (MSIT) enforces it. Fines reach KRW 30 million, for example for not giving notice or not appointing a representative, and MSIT said it would wait a year before imposing them. An amendment took effect in July 2026; check the current text before relying on the details.
Elsewhere
- China: rules for algorithmic recommendation (2022), deep synthesis (2023), generative AI services (2023); algorithm filing; mandatory labeling of AI-generated content (2025).
- Canada: the proposed AIDA died in January 2025; the federal Directive on Automated Decision-Making governs government use with algorithmic impact assessments.
- United Kingdom: principles-based, regulator-led approach; the AI Security Institute tests frontier models.
- Japan: AI Promotion Act (2025), a light-touch approach.
- Brazil: risk-based AI bill progressing.
- Singapore: voluntary frameworks and testing (AI Verify).
Check your understanding
Which law requires annual independent bias audits of automated employment decision tools?
Under Colorado's 2026 law on automated decision-making, what can a consumer ask for after an adverse consequential decision?
Under South Korea's AI Basic Act, what must a provider of a generative AI service do?