Global AI Regulation Tracker

The world's most comprehensive AI law. Classifies AI systems by risk level (unacceptable, high, limited, minimal) with corresponding obligations. Banned social scoring and real-time biometric surveillance in public spaces.

Opted for a sector-specific, pro-innovation approach rather than comprehensive legislation. Existing regulators (FCA, Ofcom, CMA) apply AI principles within their domains. The AI Safety Institute conducts frontier model evaluations.

First country to regulate generative AI specifically. Requires algorithmic registration, content labeling, and government security reviews. Generative AI services must reflect 'core socialist values.'

Part of Bill C-27 (Digital Charter Implementation Act). Would create a risk-based framework similar to the EU AI Act. Progress has been slow due to parliamentary dynamics and industry pushback.

Modeled partly on the EU AI Act with a risk-based classification system. Would establish rights for people affected by AI decisions, including explanation and human review requirements.

India has avoided binding AI-specific legislation, preferring voluntary guidelines and sector-specific advisories. Focus on promoting AI innovation and adoption rather than regulation.

Social Principles of Human-Centric AI guide voluntary industry compliance. Japan focuses on promoting innovation and has been cautious about binding AI regulation, preferring soft law approaches.

Passed the AI Basic Act in early 2025, establishing a risk-based classification system. Focuses on high-impact AI systems in healthcare, education, and employment with mandatory impact assessments.