Enforcement Actions Monitor — SEC, FCA, ASIC, CFTC & FinCEN

One unified, normalised feed of financial-regulator enforcement actions across the US, UK and Australia, refreshed on every scheduled run. Seven official regulator feeds, one consistent record shape, and only new actions are emitted — so every dataset item is an actionable event, not a re-scrape.

Sources (v1)

Source key	Regulator	Region	What it covers	Depth of feed
sec_litigation	SEC	US	Litigation releases (civil court actions)	latest ~25 releases (≈3 weeks)
sec_administrative	SEC	US	Administrative proceedings & orders	latest ~25 releases (≈2-3 weeks)
fca_news	FCA	UK	News, press releases, statements (incl. fines & bans)	latest 20 items (≈1 month)
fca_warnings	FCA	UK	Warning List — unauthorised & clone firms	latest 20 warnings
asic	ASIC	AU	Enforcement-tagged media releases	back to 2000 (full archive)
cftc	CFTC	US	Enforcement actions & orders	latest 10 releases (≈3 months)
fincen	FinCEN	US	Enforcement actions & consent orders	back to 1999 (full table, 119 actions)

Every source is a public, keyless regulator publication — no API keys, no logins, no ToS risk in your compliance chain.

Honest limits, so you can plan around them:

• The SEC, FCA and CFTC publish only their most recent items in the feeds this actor reads. A sinceDays larger than the depth column above cannot invent history the regulator does not publish there. ASIC and FinCEN do reach back years.
• SEC records carry no summary. The SEC's feeds repeat the case name in the description field, so summary is null rather than a fake echo of the title. That also means keywords matches only the case name for SEC records — filter SEC by actionTypes and party names, not by topic words like "crypto". Fetching SEC detail pages for real summaries is on the v1.1 roadmap.
• There is no keyless FCA enforcement-only feed: the FCA's filtered enforcement search sits behind bot protection. FCA enforcement reaches you through fca_news (its fines and bans are press releases) plus your keywords filter.
• parties is filled only where the regulator's title IS the party name (SEC, FinCEN, FCA Warning List). For narrative headlines (ASIC, CFTC, FCA news) parties is an empty array — a deliberately empty field beats a guessed entity in a compliance pipeline.

Example output records

An FCA Warning List entry (the highest-value record type for KYC screening):

{
  "source": "fca_warnings",
  "regulator": "FCA",
  "region": "UK",
  "sourceId": "173586|2026-08-28",
  "title": "reclaim-experts.com (Clone of FCA Authorised firm) (new)",
  "summary": "[Clone] Fraudsters copy the details of firms we authorise to try and convince people that their firm is genuine... This firm is not authorised by us but has been contacting people pretending to be an authorised firm.",
  "actionType": "warning",
  "parties": ["reclaim-experts.com"],
  "publishedDate": "2026-08-28",
  "url": "https://www.fca.org.uk/news/warnings/reclaim-expertscom-clone-fca-authorised-firm",
  "detectedAt": "2026-08-31T10:07:00Z"
}

A FinCEN consent order:

{
  "source": "fincen",
  "regulator": "FinCEN",
  "region": "US",
  "sourceId": "2026-02",
  "title": "In the Matter of UBS Financial Services Inc.",
  "summary": "FinCEN enforcement action, matter 2026-02; financial institution type: Securities and Futures",
  "actionType": "administrative",
  "parties": ["UBS Financial Services Inc."],
  "publishedDate": "2026-08-03",
  "url": "https://www.fincen.gov/system/files/2026-07/UBS-Consent-Order.pdf",
  "detectedAt": "2026-08-31T10:07:00Z"
}

Notes on the shape: sourceId is stable per source and is the dedupe key (SEC/CFTC use the regulator's own release number, FinCEN the matter number, ASIC the NN-NNNMR document number). FCA Warning List ids carry the published date, so a warning the FCA updates emits again — for a screening pipeline an updated warning is a new event. detectedAt always tells you when the monitor first saw the record.

Input options

Field	Type	Default	Description
sources	array	all seven	Which regulator feeds to monitor (keys in the table above)
actionTypes	array	[] (all)	litigation, administrative, warning, media-release
keywords	array	[]	Case-insensitive filter on title + summary; empty = everything
sinceDays	integer	14	Backfill window in days on the first run
maxItems	integer	500	Cap on records emitted per run (overflow emits next run)
userAgent	string	actor default	Declared contact string — the SEC's automated-access policy expects one. Set your own.

actionTypes maps across regulators like this: litigation = SEC litigation releases + CFTC enforcement actions; administrative = SEC administrative proceedings + FinCEN consent orders; warning = FCA Warning List; media-release = ASIC enforcement releases + FCA news.

How to use it

Schedule the actor (daily is ideal). The first run backfills the last sinceDays days; every later run emits only actions it has never seen before, so you can wire the dataset straight into Slack, Zapier/Make, or your own ingestion with no dedupe logic on your side.

Use cases

• KYC / onboarding screening — "FCA warning list API", "unauthorised firms feed": pipe fca_warnings into onboarding checks so a clone or unauthorised firm is blocked the day the FCA names it, with parties giving you the domain or trading name to match on.
• Compliance & AML monitoring — "SEC enforcement actions monitor", "FinCEN consent orders feed": one scheduled feed replaces manually checking five regulator sites; filter keywords for your counterparties, or actionTypes: ["administrative"] for orders and consent decrees only.
• Risk data & financial journalism — "cross-border regulatory enforcement data": a single normalised stream of SEC, FCA, ASIC, CFTC and FinCEN actions with regulator, region and action type already tagged, ready for a database or a newsroom alert.

Failure handling & update promise

• One regulator being down or redesigned never kills a run: that source is logged as degraded and every other source still delivers. If every source fails the run fails loudly rather than reporting a false quiet day.
• Schema drift is logged with the exact field that moved, never a crash — and ASIC's known dirty metadata values are normalised, not rejected.
• Records beyond maxItems are not marked as seen, so nothing is lost — they emit on the next run. The cap is shared out evenly between the sources that have records (with the unused remainder redistributed), so a busy regulator never starves a quiet one out of a capped run.
• Retries are spent only where they can help: a rate limit or a 5xx is retried with backoff, a moved page is not. If a regulator refuses the declared userAgent (HTTP 403) the run stops with that fix in the message rather than reporting seven dead feeds.
• Source markup and feed URLs are re-checked and test fixtures refreshed on a regular cadence; parser fixes ship as patch releases.

Pricing events

Event	Price	Charged
actor-start	$0.005	once per run
action-detected	$0.02	per NEW enforcement action emitted

Low volume by design — you pay for coverage, not bulk. A 14-day backfill across all seven sources measured 62 records on 2026-08-31; a typical quiet day is a handful of events.

Changelog

• 0.1.1 (2026-08-31) — maxItems is now split fairly across the sources that returned records (unused shares are redistributed), so a busy regulator can no longer consume the whole cap and leave the others at zero. A moved or missing page (404) fails immediately instead of being retried three times against the regulator's rate budget; only 429/5xx/timeouts are retried. A refused User-Agent (401/403) now fails the run with the exact fix in the message instead of being reported as "all sources failed".
• 0.1 (2026-08-31) — initial release: SEC litigation releases and administrative proceedings, FCA news and Warning List, ASIC enforcement media releases, CFTC enforcement actions, FinCEN consent orders. Monitor state in a named key-value store; keyword and action-type filters, backfill window, per-run item cap, per-source failure degradation.