Pyyan / Compare / UK AI Security Institute vs EU AI Office vs CAC vs Intl. Network for AI Measurement
UK AI Security Institute vs EU AI Office vs CAC vs Intl. Network for AI Measurement
4 of 5Regulators · verified 13 Aug 2026
| Specification | UK AI Security Institute | EU AI Office | CAC | Intl. Network for AI Measurement |
|---|---|---|---|---|
| Summary | Frontier model pre-deployment testing. | First hard enforcement regime for AI. | Cyberspace Administration of China. | Ten-member evaluation consortium. |
| Jurisdiction | United Kingdom | European Union | China | Multilateral |
| Instrument | Sectoral | EU AI Act | Cybersecurity Law | Shared evaluations |
| Enforcing since | 2024 | 2 Aug 2026 | 1 Jan 2026 | 2024 |
| Category | Regulators | Regulators | Regulators | Regulators |
| Official | United Kingdom ↗ | European Commission ↗ | China ↗ | Multilateral ↗ |
Highlighted rows are where these differ.
EU AI Office
- Enforcement began 2 August 2026
- High-risk obligations live the same day
- EU evaluation capacity expected 2027
Best for —.
CAC
- AI security review and data localisation
- Fines and suspensions issued without published reasoning
Intl. Network for AI Measurement
- Renamed from the Network of AI Safety Institutes
- Australia, Canada, EU, France, Japan, Kenya, Korea, Singapore, UK, US