Thread: How the World Is Regulating AI — A Comparison

1/ The world is taking radically different approaches to AI regulation. Here's where things stand — and what the differences reveal about values and power. 🧵

2/ EUROPEAN UNION — The EU AI Act is the most comprehensive AI regulation anywhere. Risk-based framework: banned uses (social scoring), high-risk (hiring, law enforcement) with strict requirements, limited-risk (chatbots) with transparency rules, minimal-risk (spam filters) with no requirements.

3/ The EU's approach: protect citizens first, innovate within guardrails. Critics say it'll slow European AI development. Supporters say it'll create trust that drives adoption. The redistribution angle: it's the only framework that explicitly limits how AI can be used against people.

4/ UNITED STATES — Fragmented approach. No comprehensive federal AI law. Executive orders set guidelines but lack enforcement teeth. Sector-specific rules emerging (FDA for medical AI, SEC for financial AI). State-level patchwork (Colorado, California leading).

5/ The US approach: let the market lead, regulate harms after they occur. This favors incumbents and fast movers. The redistribution angle: absence of regulation is itself a policy choice — it redistributes power to companies and away from affected communities.

6/ CHINA — Comprehensive but control-oriented. Regulations on recommendation algorithms, deepfakes, generative AI, and LLMs. All AI services must align with "socialist core values." Real enforcement — companies have been fined and required to modify systems.

7/ China's approach: strategic national asset, tightly controlled. The redistribution angle: regulation serves state interests, not individual rights. Workers, citizens, and communities have no independent voice in how AI is governed.

8/ UNITED KINGDOM — "Pro-innovation" framework. No new AI-specific laws. Instead, existing regulators (FCA, Ofcom, etc.) apply current rules to AI in their domains. Light-touch by design — explicitly positioned to attract AI investment post-Brexit.

9/ UK's approach: regulatory competition — be friendlier than the EU to attract companies. The redistribution angle: competing on regulatory leniency means competing to offer companies fewer obligations to the public. The public isn't the customer here.

10/ GLOBAL SOUTH — Mostly without dedicated AI regulation. Brazil advancing the most comprehensive framework. India exploring sector-specific rules. African Union has a continental AI strategy but limited enforcement capacity. The gap: countries most affected by AI labor displacement have least regulatory power.

11/ The pattern: Countries with the most AI industry regulate the least (or most favorably to industry). Countries where AI's costs land — job displacement, data extraction, environmental burden — have the least regulatory capacity. This is the global redistribution of AI governance.

12/ What would redistributive AI governance look like? International standards with real enforcement. Affected community representation in rule-making. Capacity building for Global South regulators. And a simple principle: the people affected by AI should have power over how it's governed. @redistributed

---

LinkedIn version:

The global AI regulation landscape reveals a fundamental redistribution question: who gets to set the rules for the most transformative technology of the century?

The EU AI Act takes a rights-first approach — the only framework that explicitly limits how AI can be used against people. The US takes a market-first approach — no comprehensive federal law, which itself is a policy choice that redistributes power to companies. China regulates comprehensively but for state control, not individual rights. The UK competes on regulatory leniency to attract investment.

The most revealing pattern: the Global South — where AI's costs land most heavily (job displacement in call centers, data extraction, environmental burden from data centers) — has the least regulatory capacity. Countries most affected have the least power to shape the rules.

This mirrors historical infrastructure governance. International telecommunications regulations were shaped by the countries that owned the networks. Financial regulations favor the countries that host major markets. AI governance is following the same pattern.

Redistributive AI governance would require international standards with real enforcement, affected community representation in rule-making, capacity building for Global South regulators, and a simple principle: the people affected by AI should have power over how it's governed.

The question isn't whether AI should be regulated. It's whether regulation serves the public or the industry. So far, the answer varies by geography — and the geography of regulation doesn't match the geography of impact.

#AIRegulation #EUAIAct #AIPolicy #GlobalGovernance #Redistribution