Adverse Media Watch: Negative News Screening for KYC

Use when…

• You need negative-news / adverse-media screening for a person or company (KYC).
• A compliance agent must search recent headlines and score risk from news hits.
• You want a watchlist of subjects with timespan and minimum risk filters.

Try this input

Copy-paste this JSON as the Actor input (it matches a successful run / the default schema):

{
  "subjects": [
    {
      "name": "Boeing",
      "type": "company"
    },
    {
      "name": "Sam Bankman-Fried",
      "type": "person"
    }
  ],
  "timespan": "3d",
  "minRiskScore": 20,
  "useGdelt": true,
  "useRssFeeds": true,
  "watchlistId": "default"
}

Sample output (real run 2026-10-25)

One dataset item, shortened:

{
  "key": "company:Boeing",
  "baseline": false,
  "changed": false,
  "change_count": 0,
  "changes": [],
  "change_types": [],
  "subject": "Boeing",
  "type": "company",
  "hits_in_window": 0,
  "max_risk_score": 0,
  "hits": [],
  "gdelt_available": false,
  "…": "+3 more keys"
}

What it does: Daily adverse media / negative news screening for people and companies. It searches global news (GDELT, 100+ languages of sources) plus major publisher and regulator feeds (BBC News, CNBC, Sky, NPR, US DOJ, SEC and more), scores each hit for risk and returns only new hits since the last run.

Who it's for: KYC/AML and onboarding teams, compliance and fintech agents, supplier and third-party risk, investors doing ongoing due diligence.

Price: Pay per event: $0.003 per name checked + $0.01 per new adverse media hit. Baseline runs are never charged for hits. No monthly fee.

Compliance kit

Use it with sanctions-watchlist (new OFAC/UK/EU/UN matches) and sanctions-screening (one-off checks) for sanctions plus adverse media monitoring, the two ongoing checks most AML programmes need.

Fewer false matches

• Exact phrase matching on the name or aliases (word boundaries; company suffixes like Ltd/PLC/Inc are handled). Short names are matched case-sensitively.
• context terms must also appear (e.g. a person's employer or town). exclude terms rule a hit out (e.g. a footballer namesake).
• Risk scoring 0-100 across categories: financial crime (fraud, bribery, laundering) 40, sanctions/terror 40, criminal (arrest, charged, convicted) 30, legal/regulatory (lawsuit, fined, probe) 20, insolvency 20, ESG/safety 15. minRiskScore (default 20) filters noise.
• Deduped across outlets (syndicated copies are listed in also_reported_by).

How it works

1. First run = baseline. Everything currently found is saved to your named key-value store adverse-media-watch under your watchlistId (also as the run's SNAPSHOT record). You see the current results in each row, but no new-hit charges.
2. Every later run reports only what is new since the last run (changed, changes, change_types).
3. If a source is down, that row gets an error and the old snapshot is kept, so nothing is missed or double-counted.

Quick example

Input

{"subjects": [{"name": "Acme Widgets Ltd", "type": "company"}, {"name": "John Smith", "type": "person", "context": ["Acme"], "exclude": ["football"]}], "timespan": "3d", "watchlistId": "kyc-book"}

Change text (later run)

[risk 70: criminal, financial_crime] Acme Widgets director charged with fraud (bbc.co.uk, 2026-10-22) https://...

Use it from AI agents (MCP) and code

• Apify MCP server: add https://mcp.apify.com/?tools=tudorlane/adverse-media-watch to Claude, Cursor, ChatGPT connectors, VS Code or any MCP client. The Actor appears as a tool the model can call.
• REST (sync): POST https://api.apify.com/v2/acts/tudorlane~adverse-media-watch/run-sync-get-dataset-items?token=YOUR_TOKEN with the input JSON as the body.
• Schedule it: Apify Console > Schedules > daily, with the same watchlistId. Pipe new items to Slack, email, a webhook, n8n, Make or Zapier.
• Stateless agents: pass the previous run's SNAPSHOT record as previousSnapshot instead of relying on the saved watchlist.

Sample output (real run with the default input, 25 Oct 2026)

Subject	Type	Negative hits in window	Max risk score	News search available
Boeing	company	0	0	yes
Sam Bankman-Fried	person	0	0	yes

On this run neither name had new negative news in the time window; when there are hits, each row lists the articles, risk score and why it matched.

Use cases

1. KYC/AML: negative news on customers
2. Supplier and partner due diligence
3. Ongoing monitoring of named people or companies
4. Compliance agents

Limits & failure modes

• News search, not a regulated screening database
• Matches need human review; common names can give false hits
• Coverage limited to the feeds/sources it searches
• Risk scores are heuristic from news text — not a court finding or official adverse-media database.
• Common names produce noise; tighten timespan, add identifiers, or raise minRiskScore.
• GDELT/RSS gaps mean absence of hits is not proof of a clean record.

FAQ

Is a hit a finding? No. It's a lead for a human or agent to review. Headlines and links only; full articles belong to the publishers.

Coverage? GDELT covers worldwide online news; publisher feeds add reliable UK/US business and regulator news. GDELT is rate-limited, so about 8 names are searched per request; if it's unavailable, the row shows gdelt_available: false and RSS results still come back.

Support? Open an issue on the Actor's Issues tab.

---

Keywords: adverse media screening API, negative news screening, KYC adverse media, AML monitoring, reputational risk, media monitoring API, PEP and sanctions companion, compliance automation, MCP tool, Apify Actor, AI agent tool.

Search terms: news scraper, adverse media screening, negative news search, KYC news monitoring, AML media check