top of page

● Live · Updated Apr 8, 2026

The global intelligence layer
for AI tech policy

The real gap in AI policy isn’t access to information — it’s actionable intelligence. Raw legislative text, headline news, and lengthy research reports are readily available. What’s missing is clear synthesis: what these rules actually mean in practice, where the policy landscape is heading, and what organizations should do about it. That’s exactly what this site delivers.

JURISDICTIONS TRACKED

30

across 6 regions

ACTIVE BILLS / ACTS

6

+7 this month

RESEARCH SOURCES

10

sources indexed

LAST POLICY UPDATE

11 days ago

Test-time compute: thinking, fas...

🗺

Global policy map

EU, US Federal & State, UK, China, Global South — mapped by regulatory posture

47 jurisdictions · interactive map

AI dev stage policy

Pre-training, alignment, test-time compute, deployment — rules per stage

5 stages · 80+ policy actions

🏗

Infrastructure regulation

Data centers, water use, energy, cooling, environmental compliance

14 countries · active regulation

geo

European Commission · Apr 2

EU AI Act GPAI code of practice: third draft released

New obligations for frontier model providers including incident reporting thresholds and systemic risk classification criteria.

geo

US Congress · Apr 1

Senate AI working group releases bipartisan framework draft

Proposes sector-specific oversight rather than horizontal legislation, with NIST designated as coordination body.

stage

NTIA · Mar 31

NTIA issues RFI on compute thresholds for training oversight

Seeks public comment on whether 10^26 FLOP threshold adequately captures frontier model risk levels.

infra

IEA · Mar 30

Data center electricity demand to double by 2030, IEA projects

AI workloads driving 40% of new data center capacity globally; water stress emerging as regulatory flashpoint in arid regions.

stage

DSIT · Mar 27

UK DSIT consults on mandatory safety evaluations for post-training fine-tuning

Proposal would extend frontier safety testing requirements to alignment and RLHF phases, not just base model training.

LATEST POLICY NEWS

REGIONS

🇪🇺

European Union

14 updates

🇺🇸

United States

14 updates

🇩🇪

Germany

8 updates

🇫🇷

France

7 updates

🇳🇱

Netherlands

6 updates

🇬🇧

United Kingdom

6 updates

International AI Safety Report 2026

International AI Safety Report · Jan 2026

Carbon Considerations for Large Language Model Training

Google · Apr 2021

Attention Is All You Need

Google Brain · Jun 2017

Concrete Problems in AI Safety

OpenAI / Google Brain · Jul 2016

RECENT RESEARCH

01

Pre-training

Compute thresholds, data licensing, training data provenance, copyright

High

02

Post-training / Alignment

RLHF oversight, safety evaluations, red-teaming mandates, fine-tuning rules

Active

03

Test-time Compute

Inference scaling, reasoning model behavior, agentic deployment limits

Emerging

04

Deployment & Applications

High-risk sector rules, transparency, conformity assessments, audits

Very high

05

Infrastructure

Data centers, energy, water, cooling, environmental impact assessment

Growing

AI POLICY BY DEVELOPMENT STAGE

bottom of page