Methodology
How AQ26 separates validated evidence, candidate overlays, supporting context and unredacted review material.
Evidence classes
Validated overlay: retained from the validated DEFRA/AURN overlay register. Candidate under review: monitoring candidate discovered by evidence workflow and not yet promoted. Fallback needed: OpenAQ discovery did not produce a suitable candidate; manual/alternative-source review is required.
AI/ML use
AI/ML features are transparent triage aids: score bands, candidate classes and review priority. They do not make legal, health, causal or regulatory findings.
Legal and public safety notice
All content is informational, provisional and evidence-led. The site avoids claims of breach, causation or health impact. Unreviewed data stays in the protected review site.