An independent public resource of AI policy changes from around the world.
EU AI Act Article 50 duties enter the enforcement phase
The AI Office and national authorities can enforce the Act as Article 50 transparency duties apply from 2 August 2026. Immediate compliance work covers machine-readable marking and the remaining high-risk timetable.
Article 50 applies from 2 August 2026; AI-generated or altered content is subject to machine-readable marking duties.
H.R. 9442
A federal AI data-centre moratorium enters the US debate
H.R. 9442 introduces a direct growth-control ask into Congress. Its significance lies less in near-term passage than in linking AI capacity to electricity, water and national-security constraints.
Coverage gap Coverage is concentrated in bill text, energy-trade specialist reporting, sponsor press and niche political outlets; large national center-right and business press remains sparse relative to that specialist and niche track.
IEA Energy and AI update
IEA: data-centre electricity demand grew 17% in 2025
The IEA’s Key Questions on Energy and AI executive summary reports 17% growth in data-centre electricity demand in 2025 and 50% growth for AI-focused centres, with a central projection near 950 TWh by 2030.
Coverage gap National grid-operator series can diverge from IEA global totals; measurement definitions differ across those coverage tracks.
AI-agent implementation opinions
China sets a policy direction for autonomous AI agents
New guidance frames agents as systems with perception, memory, decision and execution capabilities, pairing deployment goals with standards, certification and lifecycle governance.
Coverage gap English-language comparative coverage of standards and certification paths remains thin relative to Chinese official text.
H.R. 8094
US proposal would put foundation-model disclosure under the FTC
H.R. 8094 asks the Federal Trade Commission to establish public-information requirements covering foundation-model training data and algorithms. The vehicle remains at the introduced stage.
Coverage gap Debate coverage often collapses training-data disclosure with model-evaluation mandates; the bill text treats them as distinct information requirements.
Korea AI Basic Act
Korea’s AI Basic Act takes effect with labelling and safety duties
MSIT announced that the AI Basic Act and Enforcement Decree took effect on 22 January 2026. The package includes generative-AI labelling duties and a grace period of at least one year for most penalties.
Coverage gap Korean gazette text is authoritative; MSIT English announcement coverage does not substitute for it.
C2PA / Content Credentials
C2PA’s Content Credentials remain the open provenance baseline
C2PA Specification 2.3 defines a technical standard for content provenance and authenticity. Whether governments should mandate it is a separate ask.
Coverage gap Legal mandates that cite C2PA remain sparse relative to voluntary-adoption coverage.
California SB 53 / TFAIA
California requires large frontier developers to publish safety frameworks
SB 53, chaptered in September 2025, took effect on 1 January 2026 as Business and Professions Code Chapter 25.1. Covered large frontier developers must publish a frontier AI framework and related transparency reports.
Coverage gap Post-effective compliance-practice coverage remains thinner than chaptering and statute-text coverage.
GPAI Code of Practice
EU publishes a voluntary GPAI Code of Practice for AI Act duties
On 10 July 2025 the Commission announced receipt of the final General-Purpose AI Code of Practice. The code covers transparency and copyright for all GPAI providers and safety/security for systemic-risk models.
Coverage gap Signatory uptake and post-endorsement practice remain the live dispute after the July 2025 final text.
Japan AI Promotion Act
Japan enacts an AI promotion statute with guideline duties
Act No. 53 of 2025 frames AI-related technology as foundational and directs the national government to establish guidelines aligned with international norms.
Coverage gap Japanese gazette and guideline text remains the operative authority; English translation coverage trails that primary track.
BIS advanced computing IFR
BIS tightens EAR controls on advanced computing chips
A 16 January 2025 interim final rule revises Export Administration Regulations covering advanced computing integrated circuits used in AI datacentre contexts.
Coverage gap Partner licence practice and later amendments shape real export outcomes beyond the interim final rule text.
CoE AI Framework Convention
Council of Europe opens a binding AI human-rights treaty for signature
CETS No. 225 aims to keep AI lifecycle activities consistent with human rights, democracy and the rule of law. Entry into force still depends on the five-party threshold in Article 30.
Coverage gap Signature volume exceeds ratification; Treaty Office tallies show one ratification (EU) as of mid-2026, below the entry-into-force threshold.
OECD AI Recommendation
OECD revises its AI Recommendation and shared system definition
On 3 May 2024 the OECD Council revised the Recommendation on Artificial Intelligence. The instrument remains soft law, but its AI-system definition is widely reused in domestic drafting.
Coverage gap Domestic incorporation of the 2024 revision remains uneven across adherents.
UK AI white paper
UK keeps AI principles with existing regulators, not an early statute
The March 2023 white paper states the five AI principles will start on a non-statutory basis and be implemented by existing regulators.
Coverage gap Later legislative hardening proposals sit outside the 2023 white paper text that this coverage quotes.
NIST AI RMF
NIST keeps the AI Risk Management Framework voluntary
NIST’s AI RMF, released on 26 January 2023, remains a voluntary organisational tool. Later profiles address generative AI and critical infrastructure without converting the core framework into a statute.
Coverage gap Contractual and sectoral mandates that cite the RMF vary by sector; the framework text itself stays voluntary.
UNESCO AI ethics recommendation
UNESCO’s AI ethics recommendation remains the global soft-law baseline
In November 2021, 193 UNESCO Member States adopted the Recommendation on the Ethics of Artificial Intelligence. The live dispute is national implementation, not the fact of adoption.
Coverage gap National implementation varies widely; adoption does not equal domestic enforceability.
Singapore Model Framework
Singapore keeps AI governance on a voluntary model framework
IMDA’s Model AI Governance Framework second edition remains a voluntary ready-to-use tool for organisational AI governance.
Coverage gap Later agentic-AI factsheets extend the toolkit beyond the second-edition Model Framework.
Questions moving policy
Disputes tracked apart from the bills and headlines that carry them.