Comparing regulatory approaches
A side-by-side **orientation** to how major jurisdictions approach AI governance—useful before reading any single statute. Congressional Research Service report **R48555** (*Regulating Artificial Intelligence: U.S. and International Appr...
What it is
A side-by-side orientation to how major jurisdictions approach AI governance—useful before reading any single statute. Congressional Research Service report R48555 (*Regulating Artificial Intelligence: U.S. and International Approaches and Considerations for Congress*) summarizes diverging approaches for lawmakers.
Visual Spec & Architecture Diagram
Compare approaches: risk-based product rules vs principles vs sectoral vs agency enforcement—2×2 or table graphic.
Why it matters
Builders and citizens otherwise mash EU duties into US blogs (or vice versa). Comparison clarifies *instrument type*: horizontal regulation, sectoral enforcement, voluntary frameworks, principles, treaties.
How it works (plain) — CRS-aligned high level
| Approach | Sketch (verify in sources) |
|---|---|
| EU | Broad, horizontal, risk-based AI Act with tiered obligations (Commission: force date 1 Aug 2024) |
| United States | No single comprehensive federal AI statute in the CRS framing; sectoral agency authorities, voluntary frameworks (NIST RMF), federal-use guidance, state activity, executive plans |
| United Kingdom | Measured / sector-regulator-forward approach (CRS pairs UK with “measured” peers) |
| China | Targeted rules + strong state role in development/security goals; often described as more vertical/technology-specific (CRS) |
| International soft/hard law | OECD principles; UNESCO ethics Recommendation; CoE Framework Convention treaty |
Debates CRS surfaces: whether broad federal rules reduce uncertainty vs whether they risk slowing innovation amid geopolitical competition—present both sides as policy arguments, not truths.
Everyday example
A “compliance spreadsheet” with one column labeled “AI law” fails; you need columns per market and instrument type.
Try it
Using only primary links in Sources, write one sentence each for EU vs US vs OECD distinguishing *binding product regulation* vs *voluntary risk framework* vs *principles recommendation*.
Myths
- ⚠️ Myth: The strictest region’s blog rules the world automatically.
- ✓ Reality: Reach depends on where you place systems/serve users—counsel must map facts.
- ⚠️ Myth: Comparative charts never go stale.
- ✓ Reality: Re-verify quarterly; this page’s
last_verifiedis a reminder.
Sources
- CRS R48555: https://www.congress.gov/crs-product/R48555 ↗
- EU Commission AI Act force news: https://commission.europa.eu/news-and-media/news/ai-act-enters-force-2024-08-01_en ↗
- NIST AI RMF 1.0: https://www.nist.gov/publications/artificial-intelligence-risk-management-framework-ai-rmf-10 ↗
- OECD Principles: https://legalinstruments.oecd.org/public/doc/648/dd63ee37-eef0-40d8-9480-26c011db227d.htm ↗
- UNESCO: https://www.unesco.org/en/artificial-intelligence/recommendation-ethics ↗
- CoE Convention hub: https://www.coe.int/en/web/artificial-intelligence/the-framework-convention-on-artificial-intelligence ↗
- White House AI Action Plan: https://www.whitehouse.gov/releases/2025/07/white-house-unveils-americas-ai-action-plan/ ↗
