🤖 AI Tools
· 3 min read

AI Data Privacy Laws by Region — US, EU, UK, China, India (2026)


If your AI product has users in multiple countries, you’re subject to multiple privacy regimes. Here’s what developers need to know for each region.

Quick comparison

RegionLawAI-specific?Max fineData residency required?
EUGDPR + AI Act✅ AI Act7% revenueNo (but transfers restricted)
US (California)CCPA/CPRA❌ General$7,500/violationNo
UKUK GDPR + AI framework⚠️ Guidance only£17.5M or 4%No
ChinaPIPL + AI regulations✅ Multiple¥50M or 5%✅ Yes (for certain data)
IndiaDPDP Act 2023❌ General₹250 crore (~$30M)⚠️ Restricted transfers

European Union

The strictest regime. Two laws apply:

GDPR — Covers all personal data processing. Requires consent or legitimate interest, DPAs with processors, and transfer mechanisms for data leaving the EU. See our GDPR guide for developers.

EU AI Act — Risk-based AI regulation. Full enforcement August 2026. High-risk AI systems need documentation, testing, and human oversight.

For developers: Use EU-based providers like Mistral or self-host to simplify compliance.

United States (California)

No federal AI law yet. California’s CCPA/CPRA is the strongest state-level regulation:

  • Right to know what data is collected
  • Right to delete personal data
  • Right to opt out of data “sales” (including sharing with AI providers)
  • Automated decision-making transparency requirements

For developers: If you send user data to AI APIs, disclose it in your privacy policy. Allow opt-out. California’s definition of “sale” is broad enough to cover sending data to third-party AI providers.

United Kingdom

Post-Brexit, the UK has its own GDPR (nearly identical to EU GDPR) plus a principles-based AI framework:

  • No AI-specific legislation (yet)
  • Sector regulators (FCA, Ofcom, etc.) apply existing rules to AI
  • ICO guidance on AI and data protection
  • Adequacy decision with EU (data can flow freely)

For developers: Treat UK like EU for data protection. The AI framework is guidance, not law — but regulators are watching.

China

The most complex regime for AI:

  • PIPL (Personal Information Protection Law) — China’s GDPR equivalent
  • AI regulations — Specific rules for generative AI, deepfakes, recommendation algorithms
  • Data localization — Certain data must stay in China
  • Algorithm registration — AI services must register with the government

For developers: If you serve Chinese users, data likely must stay in China. Using Chinese AI providers (DeepSeek, Qwen, Kimi) simplifies this but adds complexity for non-Chinese users.

India

The Digital Personal Data Protection (DPDP) Act 2023:

  • Consent-based framework
  • Data localization for “critical” personal data
  • Right to erasure
  • Penalties up to ₹250 crore (~$30M)
  • Still being implemented — rules expected through 2026

For developers: Similar to GDPR in spirit but less prescriptive. Watch for implementation rules.

Practical advice for global products

  1. Default to GDPR compliance — it’s the strictest, and meeting it usually satisfies other regimes
  2. Use self-hosted models for maximum flexibility across regions
  3. Offer data residency options if serving EU and China
  4. Document everything — every regime requires demonstrable compliance
  5. Use Mistral for EU, local providers for China, any provider for US/UK

Related: AI and GDPR for Developers · EU AI Act for Developers · Which AI APIs Are GDPR Compliant? · Self-Hosted AI for GDPR