The signal arrives before the article
Complaints, allegations and community warnings circulate long before a journalist writes them up. A product that only reads newspapers is reading the confirmation, not the warning.
Product · Adverse media
Adverse Media Screening & Monitoring
Screen across news, social media, court records, regulatory notices, then decide what each source is worth. Social coverage surfaces the allegation weeks before the article, weighted as a signal rather than as evidence.
All Languages
covered
Social Media
WHERE THE STORY STARTS
Real-Time
Monitoring interval
Customizable
Choose your keywords, sources and categories
Why this matters
There is no official list that tells you what counts as adverse.
Sanctions give a definitive answer: sanctioned or not. Adverse media gives you a judgment, and the judgment is yours to defend. A ten-year-old allegation, a founder named in someone else's case, a thread accusing a merchant of running a scam. The last one is often months ahead of the reporting, and most tools do not look there at all. Which of these belongs in your file is a policy question, and most tools answer it for you without showing their working.
No authority defines the boundary
What counts as adverse depends on your sector, your risk appetite and your regulator. It totally depends on your risk appetite.
So the vendor decides instead
A fixed relevance model, tuned for someone else's portfolio, determines what reaches your analysts and what never does.
You get noise or silence, and cannot fix either
Too broad and the queue fills with namesakes and parking fines. Too narrow, and the story that mattered was filtered out before anyone saw it.
How it works
Scroll to advance
Four steps, from a name to a decision you can defend
-
Step 01
Set what adverse means for you
Choose the offence categories in scope, the minimum source tier, how far back a story still counts, and whether allegations count differently from convictions. This is the policy, written once and applied to every screen.
-
Step 02
Search wider than the newsroom
The name is searched across news, regulatory notices, court records, social platforms, forums and local-language outlets in 30+ languages. Early signals rarely break only in English, and rarely break in a newspaper first.
-
Step 03
Resolve, classify, filter
Namesakes are separated using the identifiers you already hold. Each remaining story is read, classified by offense and source tier, and measured against your policy. What does not meet it is not deleted; it is set aside and fully auditable.
-
Step 04
Keep watching, alert on change
Monitoring runs on the same policy and rules after onboarding. A new story that meets it raises an alert and contributes to the customer risk score. A story that never reaches the queue.
complead / screen
- Retail policyv2.1
- CATEGORIESfraud, corruption, ML, trafficking
- SOURCE TIERtier 1-3
- LOOKBACK7 years
- ALLEGATIONcounts at 50%
complead / filter
- Delta Corp47 raw hits
- NEWS18
- SOCIAL AND FORUMS21
- COURT AND REGULATORY8
- LANGUAGESEN, TR, RU
complead / categorise
- 47 hits4 relevant
- DROPPED, NAMESAKE31
- DROPPED, OUT OF SCOPE12
- RELEVANT4
- SET ASIDE, REVIEWABLE43
complead / monitor
- NEW TODAY3
- Delta Corp, fraud+18 to score
- Case #4102below threshold
- Case #7719escalated
Set what adverse
Scan
Case Management
Ongoing Monitoring
Capabilities
What you can tune, and what happens when you do
Coverage beyond the newsroom
News, wire services, social media, regulatory notices, enforcement actions and court records in 30+ languages, each rated by tier, with original text retained.
Social and open web, tiered separately
Platforms, forums, community channels and blogs are searched alongside formal sources but scored as their own tier. A cluster of complaints about a merchant, an allegation circulating in a crypto community, a post the subject made themselves. Set what weight, if any, this tier carries in your policy.
Relevance as a rule, not a black box
Offence categories, source tiers, lookback period, allegation weighting and entity scope are settings on your side. Change one and the effect on your open alerts is visible before it applies. Run separate policies per segment or jurisdiction.
AI reads, your policy decides
Every item is read and classified by offense, role and severity, with the reasoning shown. On social sources, the model also separates volume from substance: fifty accounts repeating one claim is one signal, not fifty.
Namesake resolution
Identifiers you already hold separate your customer from everyone who shares their name. On social sources, handle history and profile data are used the same way.
Evidence that survives deletion
The original post or article, its translation, its source tier, its capture date and the policy version in force are retained together. Posts get deleted. Your evidence does not.
Product tour
Four screens, and one of them is the one your vendor usually keeps
Where you decide what counts
Categories, source tiers, lookback and weighting, with the effect on your current portfolio shown as you change them.
- Offence categories in and out of scope
- Minimum source tier, per layer
- Lookback period and allegation weighting
- Open alert count updating as you change it
complead / results
- Retail policyv2.1
- FRAUDin scope
- CORRUPTIONin scope
- TAXout of scope
- LOOKBACK7 years
- OPEN ALERTS341 → 288
Signal, not a news dump
What passed your policy, ranked by severity and source tier, namesakes already separated.
- Source layer shown on every hit
- Ranked by severity and tier
- Set-aside coverage still reachable
- One click to the original
complead / article-detail
- DELTA CORP 4 relevant
- TIER 1, COURT bribery probe, 2024
- TIER 2, PRESS fraud claim, 2023
- TIER 3, FORUM scam allegation, 2026
- SET ASIDE 43 reviewable
The story, and why it counted
Original text, translation, classification, and the specific policy rule that puts it in scope.
- Source text and translation together
- Predicate offense and stage tagged Source tier shown
- The rule that matched, named
complead / monitoring-feed
| ``` | |
| Delta Corp named in bribery probe | |
| SOURCE | tier 1, TR original |
| CATEGORY | corruption |
| STAGE | allegation, 50% |
| MATCHED RULE | corruption, tier 1-3 |
| ``` |
What broke since onboarding
New coverage on live customers, its contribution to the score, one action to escalate.
- Delta only, new coverage
- Contribution to the risk score
- Below-threshold items still visible
- Escalate to a case in one action
- NEW TODAY 3
- DELTA CORP, FRAUD +18
- #4102 below threshold
- #7719 escalated to case
Data coverage
Three source layers, weighted differently because they are different
A court filing and an anonymous post are not the same kind of fact, and treating them the same is what makes adverse media either useless or dangerous. Complead separates sources into layers and lets you set what each layer is worth, so breadth of coverage does not force you into a lower standard of proof.
Source and category list
Languages
Formal records
Court filings, enforcement notices, regulatory actions, disqualification registers. Verifiable, dated, attributable.
Sources
Media Reporting
International and regional outlets in 30+ languages, tiered by editorial standard, original text kept alongside translation.
Categories
Social and open web
Platforms, forums, community channels and blogs. Fastest layer, lowest verification. Configure it as an alert trigger, a score contribution, or neither.
Runs on Fusion
Fusion is what turns a story into a weight
A relevant article on its own is a document. On Fusion it becomes a scored factor on the customer record, weighted by offence severity and source tier, feeding the same risk rating that screening and behavior feed. And the rating comes back the other way: a high-risk customer gets a wider lookback and a lower alert threshold, without a second set of rules to maintain.
Empower your Unified Risk Score
With Fusion AI, you can directly bring the Unified Risk Score to decision,
This product
Client Screening & Ongoing Monitoring
Merge Adverse Media insights with 3000+ sources fed into Client Screening and Monitoring Results
Feeds into
KYB & Company Verification
Use Adverse Media results into KYB product with Fusion AI
You get
Customer Risk Assessment
Negative news is also affecting Customer Risk Score, so Fusion can do that with fully automated
Integration
One call to screen, a webhook for everything after
Send the name and the policy to apply, get classified hits back with their evidence. Subscribe to new coverage as it appears.
// screen against a policy
POST /v1/adverse-media "name": "Delta Corp Ltd", "policy": "retail_v2", "identifiers": {...}
200 OK · 4 relevant · 43 set aside · 980 ms
```
200 OK · 2 relevant · 1,240 ms
// subscribe to new coverage
```
PUT /v1/adverse-media/monitoring/8841 "events": ["hit.new","policy.published"], "url": "https://you.example/media-alert"
```
Case study · Payments
Read the case study
A payments firm caught a corruption story its old tool never saw
The customer was clean in English coverage. A local-language investigation, categorised as corruption, surfaced in the multi-language screen and the account was reviewed before exposure grew.
30+
Languages searched
2
Relevant of 40 hits
0
Namesakes in the result
See what the press says about your own names
Bring a sample. In 30 minutes you will see the coverage, the categories and the evidence a screen produces.
300,000+
Daily Articles crawled
190+
Countries covered
Real-Time Media Check
Always real-time checking against to Media & Social Media
FAQ
Questions analysts ask us
What situations count as adverse media and who is it that decides?
You do. The offence categories included within the scope, the minimum source tier, how far back a story remains valid and whether an allegation is given the same weight as a conviction are settings on your end which you set once as a policy and then apply to all screens. The vendor does not provide any default settings determining which stories reach your analysts.
Actually, do you really go through social media and forums?
Yes, together with news, wire service reports, regulatory notices and court records, social sources are searched as a separate layer and are assessed individually since a community complaint and a court filing are not of the same kind of fact. You determine the value of that layer: whether it should trigger an alert, contribute to the score, or make no difference at all.
What becomes of the coverage which fails to comply with our policy?
It is left aside rather than being deleted. All the items that the screen identified remain accessible for the reason why they were excluded, so an examiner who asks what you had seen and decided not to act on is given an answer instead of facing a gap.
How do you manage to distinguish our customer from another person who has the same name?
We use the identifiers you already hold and apply them before any item enters the queue; on social sources the handling of history and profile data is done in the same way.
What number of languages do you deal with?
Over 30, comprising regional and local-language outlets. The original text is listed alongside the translation, so a decision regarding a translated story can still be verified against what was actually published.
How do you prevent the queue being filled with the same story fifty times?
The model treats volume and substance as separate things; a single signal comes from fifty accounts making the same claim, not from fifty individual ones, and when a particular event is covered repeatedly, the coverage is grouped rather than counted each time.
Is an accusation to be regarded as equivalent to a conviction?
That is for you to decide. A tag indicating the stage is attached to each item and reflects the weight that you have set, so an unproven allegation from six years ago and a current conviction do not result in the same score.
What is the result when a post or an article is deleted?
The text captured, its translation, the source tier, the date of capture and the version of the policy in effect at that time are all kept together so that the evidence relating to a previous decision remains even if the source is no longer available.
Can we use different policies for different segments?
Yes, separate policies can be had for retail, corporate, and high risk portfolios, and altering one will show the effect on your open alerts before it is applied.
What effect does a story have on the customer's risk score?
When an item is relevant it becomes a scored factor on the customer record, the factor being weighted according to the severity of the offence and the source tier, and this contributes to the same rating that is derived from screening and behaviour. The rating also works in the reverse direction since a high risk customer is given a longer lookback period and a lower alert threshold, there being no separate set of rules to keep things going.
Testimonials
All case studies
What compliance teams say
We went from checking customers one by one to a single platform that screens more than 3,000 of them around the clock, so our analysts can finally focus on real risk.
We moved from screening customers one by one to a unified platform where our analysts can focus on what actually matters.
We focus on real risks, not false positives, meeting our AML obligations and our customers' expectations.