Methodology

How data is collected

RegWatch reads directly from official sources — regulator databases and APIs (such as openFDA), clinical trial registries (ClinicalTrials.gov), and biomedical literature indexes (PubMed). For a small number of authorities without a public API, we use targeted, scheduled web checks against the authority's own published pages rather than a third-party aggregator. A scheduled sweep re-checks each monitored molecule and records only items that are new since the last check, deduplicated by source URL.

How AI is used, and disclosed

An AI model may generate a short relevance note summarizing why an item was surfaced, and helps classify items into the Regulatory / Trial / Research / News types. AI is not used to invent facts, dates, or sources — every item you see is backed by a real source record with a real link. Where a source gives only a partial date (a year, or a year and month — common with PubMed and ClinicalTrials.gov), we show that partial date rather than fabricate a specific day.

Verification tiers

Every item carries one of four tiers: Official (a direct regulator-domain source), Pharmacist-verified (reviewed by a human), Trade press (real, but second-hand), or Rumour (explicitly unconfirmed). RegWatch never asserts something as settled fact — the linked source is the authority, not our summary of it.

Every alert links to its source

Every item includes a direct link to the original document or listing. We also check those links periodically and flag ones that have gone dead, so a broken citation doesn't sit unnoticed in your feed.

Known limitations

• Coverage is deliberately scoped, not exhaustive — see /coverage for exactly which authorities are live today versus in progress.
• Some sources are not reachable from our current hosting environment (for example, general news aggregation). When that happens, we show zero results rather than fabricate a result — an honest gap, not a failure we hide.
• RegWatch does not file, submit, or take regulatory action on your behalf. It surfaces candidate changes for your team to assess.