Methodology — AI Policy Pulse

Why this exists

What we publish

In scope

• Laws and bills with real legislative momentum
• Regulatory rules and guidance with binding or practical weight for operators
• Enforcement actions and official investigations in the public record
• Government and intergovernmental frameworks with formal status
• Court decisions that create or clarify precedent for AI-related obligations

Out of scope

• Academic papers without a regulatory or enforcement hook
• Voluntary industry pledges that are not backed by a formal instrument
• Individual company AI policies, unless elevated by statute or regulator action
• Opinion pieces and general commentary
• Press releases that are not tied to a specific, citable rule, bill, or filing

Coverage

How we verify

Every entry passes two independent reads of the source before it can publish. Neither read sees the other's work. We then compare what they drew out. When they agree on the core facts — jurisdiction, status, and what the instrument actually requires — the entry can move forward. When they disagree, an editor decides before anything goes live.

How each entry is labeled

• Auto-sourced — verified end-to-end by the system without an extra editorial pass. Used only when the source is an authoritative primary record (official gazettes, federal registers, parliamentary databases) and the two reads converge cleanly.
• Editorially reviewed — verified the same way, then checked by a human editor before publication. Used for borderline calls, novel regulatory developments, and anything where automated review surfaces uncertainty.
• Contributor submitted — surfaced by a practitioner through our submission form, run through the same verification path, and released only after editorial review.

How we score urgency

• Highest urgency. Active enforcement, near-term compliance deadlines, or rules that already bite in practice. If you are responsible for running a program, this is where you act, not just read.
• Watching. Bills moving, rules with effective dates further out, regulators signalling concrete timelines worth planning against — material, but not necessarily immediate.
• Informational. Early-stage proposals, voluntary frameworks, and official signals about future direction — worth knowing, usually not yet worth reorganizing a roadmap over.

Tier assignments can change as situations develop. When they do, we record the shift and why in that entry's history so the trail stays honest over time.

Sources we don't read in English

What we don't claim

• We're not exhaustive. We cover what we can verify from authoritative public channels. Informal pressure, off-the-record guidance, and enforcement that has not reached the open record will not appear here until it does.
• We're not real-time. We monitor on a weekly cadence. Something that posts mid-week may not surface on the site until the next review cycle — sometimes later if it needs editorial follow-up. For breaking headlines, use a wire service; use us for verified instruments.
• We don't predict. We track what is published and defensible today. Urgency tiers describe the trajectory of an instrument as we understand it from the record — not our forecast of politics or litigation.
• We will sometimes be wrong. Verification narrows error; it does not erase it. When you find a mistake, tell us — we want to fix it in the record, not quietly patch it.

Corrections