Adverse media screening is the systematic review of public reporting for information that bears on a counterparty's integrity. Regulators expect it as part of customer due diligence. The difficulty is that satisfying the expectation and producing useful intelligence are not the same exercise.
A screening programme that returns thousands of unreviewed hits is not a control.
It is documentation of an intention.
How Screening Actually Works
The mechanics are straightforward: name and entity queries run against news archives, regulatory publications and court reporting, filtered for relevance and reviewed by an analyst.
The difficulty is in every step of that sentence. Name matching across transliterations produces both false positives and silent misses. Relevance filtering discards signal alongside noise. And analyst review is where most programmes are under-resourced.
Where Automated Tools Fall Short
Automation is necessary at scale. It is not sufficient.
- Language Coverage
Most tools weight English-language sources heavily. Material reporting frequently appears first in local media. - Name Variation
Transliteration, patronymics and legal name changes defeat exact matching, and fuzzy matching floods the queue. - Context Loss
An article naming someone as a witness reads identically to one naming them as a defendant unless a human reads it. - Recency Bias
Archives thin out quickly. Older conduct that remains material may simply not be indexed.
What a Defensible Programme Requires
Documented methodology, defined escalation thresholds, analyst review of anything above them, and a retained record of what was reviewed and why it was cleared.
The test is not whether the programme found something. It is whether, if something is later discovered, the organisation can demonstrate that its process was reasonable at the time.
