HaltState.ai Triage Protect your first action
Official sources. Reviewed controls. Explicit limits.

Global AI Agent Governance and Runtime Controls

HaltState is an independent control layer for actions routed through an active governance integration. Within that boundary, it can return ALLOW, APPROVAL_REQUIRED, or DENY before a customer-integrated worker performs a protected business side effect.

Worker behavior, reporting, evidence, retries, failure handling, and stop signals remain subject to the catalogue's capability statuses and limitations. These mappings describe technical controls and evidence links that can support governance work. They are not legal advice, certification, regulator approval, or a claim that HaltState alone establishes compliance.

Active lawScheduled lawOfficial guidanceVoluntary frameworkProposed lawPolicy signal
Discuss governance readiness Open the Runtime Control Catalogue
Verified governance signals

Runtime control, mapped globally.

Explore official laws, guidance, and frameworks mapped to reviewed HaltState controls. Select a market to open its published readiness path.

Open market directory

4 published readiness markets / verified sources

Primary-source directory

Official status and last verification

Only records marked verified after direct government, regulator, or standards-body source review render here.

JurisdictionSourceOfficial statusEffectiveLast verifiedPrimary source
AustraliaAPP 1 automated-decision transparency amendmentsScheduled law2026-12-102026-07-27Privacy and Other Legislation Amendment Act 2024 OAIC APP 1 automated-decision guidance OAIC consultation on detailed automated-decision transparency guidance
AustraliaGuidance for AI AdoptionOfficial guidanceNot applicable2026-07-27Australian Government Guidance for AI Adoption implementation guidance Department publication history for the Guidance for AI Adoption
CanadaGuide on the Use of Agentic Artificial IntelligenceOfficial guidanceNot applicable2026-07-27Government of Canada agentic AI guide (official French text) Government of Canada policies and guidelines directory
European UnionAI Act Article 50 transparency obligationsScheduled law2026-08-022026-07-27European Commission Article 50 guidelines Regulation (EU) 2024/1689, Article 50
SingaporeModel AI Governance Framework for Agentic AIVoluntary frameworkNot applicable2026-07-26IMDA launch announcement IMDA May 2026 framework update IMDA Agentic AI framework (PDF)
SingaporeSafeguards for Agentic Finance at RuntimeVoluntary frameworkNot applicable2026-07-27MAS Safeguards for Agentic Finance at Runtime paper Singapore Government Press Centre MAS publication announcement
United StatesAI Agent Standards InitiativePolicy signalNot applicable2026-07-26NIST AI Agent Standards Initiative
Priority readiness paths

Take the reviewed control model into a deployment review.

EU Article 50 readinessSingapore runtime readinessCanada agentic AI control mappingAustralia ADM evidence readiness

Readiness reviews identify technical controls, customer-owned integration work, and evidence gaps. They do not provide legal certification.

HaltState control families

What catalogued controls enforce or support within their stated boundaries

Inspect stable IDs, enforcement points, evidence links, customer responsibilities, and limitations.

Source and mapping method

Trace each mapping from official source to control and evidence links.

Each public source has an official status, publication date, effective date where applicable, direct URL, and last-verified date. Each mapped HaltState control records an enforcement point, evidence links, capability status, customer responsibility, and stated limitation.

The directory includes only records marked verified after direct official-source review; review-required records are withheld. Missing HaltState capabilities remain gaps or customer-owned integration work.

Proof Packs Live evidence Technical docs Action-risk triage AI containment analysis