Adverse Media Screening: What It Is, What Regulators Expect, and What Gets You Cited
Adverse media screening is a KYC control that systematically searches news databases, court records, and other public sources for negative information about a customer or counterparty. FATF Recommendation 10 and the EU's Sixth Anti-Money Laundering Directive require institutions to run this check at onboarding and on a continuing basis throughout the customer lifecycle.
What is Adverse Media Screening?
Adverse media screening (also called negative news screening) is the process of searching publicly available sources for content that indicates a customer, beneficial owner, or counterparty poses a financial crime risk. Sources include news archives, court filings, regulatory enforcement registers, insolvency databases, and social media. The goal is to identify individuals or entities connected to money laundering, fraud, terrorism financing, corruption, or serious criminal activity before or during a business relationship.
In the Know Your Customer (KYC) stack, adverse media sits alongside sanctions screening and PEP screening. Unlike sanctions lists, which are relatively static and binary, adverse media draws from an unstructured, constantly changing body of content. That's what makes calibration harder and automation more valuable.
The scope goes beyond a basic Google search. Institutions are expected to query structured databases (LexisNexis, Refinitiv World-Check, Dow Jones Risk and Compliance), court judgment repositories, insolvency records, and regulatory enforcement registers. For higher-risk customers, dark web monitoring is increasingly common.
Adverse media screening applies at three points in the customer lifecycle: onboarding (before a relationship begins), periodic review (annual or risk-tiered), and event-triggered refresh (activated by transaction alerts, SAR filings, or significant news events). For politically exposed persons and high-risk business accounts, continuous real-time screening is now the baseline expectation across most major jurisdictions.
Why is Adverse Media Screening required?
The regulatory mandate for adverse media screening comes from several directions simultaneously.
FATF Recommendation 10 requires financial institutions to conduct customer due diligence on an ongoing basis. FATF's 2022 guidance on beneficial ownership and its updated guidance on politically exposed persons both explicitly extend that obligation to monitoring publicly available information about customers. Adverse media is no longer an implied good practice. It's referenced directly in supervisory expectations.
In the EU, the Sixth Anti-Money Laundering Directive and the European Banking Authority's 2022 Risk Factor Guidelines (EBA/GL/2021/02) make the obligation concrete. The EBA's AML/CFT guidance lists adverse media as a specific risk indicator across multiple customer and product categories. Institutions offering correspondent banking, private banking, or high-value services face the most explicit requirements.
In the US, the Bank Secrecy Act and FinCEN's Customer Due Diligence Rule (31 CFR 1010.230) don't use the term "adverse media" explicitly. But OCC examination guidance and the Federal Reserve's BSA/AML examination manual both make clear that institutions are expected to consult publicly available information when assessing customer risk. Relying on official sanctions lists alone isn't sufficient.
In the UK, the Money Laundering, Terrorist Financing and Transfer of Funds Regulations 2017 require ongoing monitoring that includes assessment of publicly available information. FCA supervisory letters to correspondent banking and wealth management firms have reinforced this expectation repeatedly.
The pattern is consistent across jurisdictions: regulators expect institutions to look at what's publicly known about their customers. Enforcement actions involving Danske Bank and Deutsche Bank both included findings that basic public-source checks weren't being conducted consistently for higher-risk populations.
What do regulators expect to see?
On exam day, examiners want evidence, not policy statements. Here's what they actually request.
Written policies and procedures. A documented policy defining what adverse media screening covers, which customers are in scope, what sources are checked, and who owns the program. The policy must specify trigger events for out-of-cycle screening and define escalation paths when a hit is detected.
Source coverage documentation. A list of every database, news feed, and public source used, with rationale for each. Examiners have cited institutions that rely on a single vendor with narrow geographic or language coverage. If you serve customers across 40 countries, your screening sources need to match that footprint.
Calibration and tuning records. Documentation showing how the system's sensitivity was set, when it was last reviewed, and what testing validated those settings. This includes false-positive rate analysis. If your tuning dismisses 99% of hits automatically, you need documented evidence showing those dismissals were justified.
Escalation and decision trails. For every hit reviewed, a record of who reviewed it, what was decided, and why. Verbal reviews don't satisfy this. Examiners want analyst reasoning in writing, not just a "cleared" checkbox.
Governance and oversight evidence. Board or senior management reporting on adverse media volumes, escalations, and program changes. MI packs with trend data. Board minutes showing the program was discussed and challenged.
Training records. Evidence that analysts have been trained on what constitutes a genuine hit, how to assess materiality, and when to escalate to the MLRO.
SLA and backlog data. Examiners ask for review turnaround times and current queue sizes. An adverse media backlog of 3,000 unreviewed alerts is a finding before they've looked at anything else.
What does good Adverse Media Screening look like?
Good adverse media screening is a program, not a checkbox. The Wolfsberg Group's Financial Crime Compliance Guidance and FATF's risk-based approach documentation both describe the components expected of mature programs. Here's what good actually looks like:
Risk-tiered coverage. All customers get a baseline screen at onboarding. High-risk customers, PEPs, and accounts flagged by transaction monitoring get continuous or quarterly screening. Low-risk retail customers get annual periodic review.
Multi-source, multi-language coverage. Wolfsberg's correspondent banking guidance specifically recommends checking local-language sources in markets where customers operate. Running English-only searches for a customer with operations in Turkey or Brazil is a documented gap that examiners will find.
Structured hit review workflow. Each hit is categorized by crime type, source credibility, and recency. A BBC article published last month is assessed differently from a regional blog post dated a decade ago.
Documented materiality thresholds. Not every mention of "fraud" is an automatic derisking trigger. Good programs have written thresholds: what content types require enhanced review, what triggers a SAR referral, and what types are dismissed with analyst rationale.
Feedback into CDD. When adverse media changes a customer's risk rating, that change is pushed into the customer due diligence record and triggers a proportionate relationship review.
Control testing. Periodic testing using known-bad test records to verify detection rates. The FATF Risk-Based Approach Guidance for the Banking Sector is the most widely cited public benchmark for testing methodology.
Vendor due diligence. If you use Refinitiv, LexisNexis, or Dow Jones, your documented due diligence on their coverage scope, update frequency, and error rates is part of your own control evidence.
Common audit findings and exam citations
The pattern of adverse media failures in enforcement actions is depressingly consistent. Institutions get cited for the same things, repeatedly.
Tuning gaps. Screening systems calibrated for low alert volume, with no documented testing of the resulting false-negative rate. OCC examination findings from 2021 and 2022 repeatedly cited institutions that could demonstrate their false-positive rate but had no data on what their settings were missing.
Onboarding-only screening. Running adverse media once at account opening and never again. The Danske Bank case is the clearest example at scale: thousands of non-resident customers with publicly documented criminal histories were maintained for years with no evidence of ongoing screening. The Estonian Financial Intelligence Unit found that adverse media controls were effectively non-functional for this population.
Alert backlogs. The FCA's 2021 supervisory review of correspondent banking flagged institutions where the median adverse media review time exceeded 90 days. A queue that old isn't screening. It's archiving.
No documentation of dismissed hits. Analysts clearing alerts without written rationale. Examiners treat undocumented dismissals as evidence the control was bypassed rather than applied.
Narrow source coverage. Checking one or two English-language aggregators for customers operating in emerging markets. The HSBC 2012 consent order identified systematic failures in due diligence on higher-risk customers, including insufficient use of available public-source information on customers whose risk profiles warranted deeper checks.
Weak governance. No senior management reporting on adverse media as a standalone control metric. No evidence the MLRO receives data on hit rates or escalations. The Deutsche Bank 2017 enforcement action included findings on risk management failures traceable to inadequate second-line oversight of front-line due diligence controls.
Metrics and KPIs
Measuring adverse media screening health requires tracking both volume and quality, separately.
Alert volume by customer tier. Total alerts generated per month, segmented by risk tier. A sudden drop in high-risk customer alerts without a corresponding drop in the population is a tuning failure signal, not a success.
False-positive rate. The percentage of reviewed alerts dismissed as non-matches. Rates above 95% in any category typically indicate settings generating noise without value. Rates below 80% may indicate genuine sensitivity gaps. Neither threshold is universal. Institutions should track their own rate over time and document any significant changes.
True-positive conversion rate. The percentage of reviewed hits resulting in escalation, enhanced review, or a SAR referral. If a program runs 10,000 reviews per month and generates zero escalations, something's wrong with either the population or the review process. This is the most direct measure of whether the program is detecting real risk.
Review SLA compliance. The percentage of alerts reviewed within defined timeframes: typically 5 business days for standard reviews, 24-48 hours for urgent or high-risk flags. Track SLA compliance separately by customer tier to see where the program is stressed.
Backlog, absolute count and trend. Total unreviewed alerts at month-end, and whether that number is growing or falling. A growing backlog is a program in distress. Publish this figure to senior management monthly.
Source coverage refresh rate. How frequently each data provider updates its records. This should be contractually defined and tracked, particularly for providers covering emerging market jurisdictions.
Periodic review completion rate. The percentage of customers due for scheduled adverse media refresh who received it on time. Distinct from alert volume, this measures whether the systematic screening program is actually executing as designed.
How Adverse Media Screening connects to other controls
Adverse media screening doesn't stand alone. Its findings feed into and draw from adjacent controls across the KYC and AML stack.
The closest relationship is with PEP screening. PEP status is often first identified through news sources rather than commercial PEP databases, which lag reality for local and regional figures. A public official in a high-corruption jurisdiction may not appear on any commercial list, but adverse media will surface coverage of their role. Many institutions run PEP and adverse media reviews with the same team, the same workflow, and shared escalation paths.
Enhanced due diligence is frequently triggered by adverse media findings. When screening surfaces a current criminal investigation or regulatory action involving a customer, the proportionate response is a full EDD refresh and a reassessment of the business relationship.
Transaction monitoring and adverse media work as cross-referencing controls. A transaction alert on an account, combined with a recent adverse media hit on the same entity, is a materially stronger escalation signal than either finding in isolation. Both controls should share context in real time.
The typologies where adverse media is most valuable include layering schemes, where corporate controllers have criminal backgrounds that surface in public records before they appear on sanctions lists, and money mule networks, where recruiting activity or prior fraud convictions appear in regional news months before law enforcement databases are updated. Adverse media is one of the few controls that can surface real criminal activity before it becomes official.
How FluxForce supports Adverse Media Screening
FluxForce's AI agents monitor adverse media signals continuously across multiple languages and source types. When a hit is detected, the relevant agent captures a complete evidence package: source, content, timestamp, and risk classification. That package is audit-ready from the moment it's created, with no manual assembly required.
Review workflows route alerts by customer risk tier, with configurable autonomy settings that let compliance teams define exactly where human sign-off is required. For teams managing high alert volumes, the backlog reduction is immediate. Book a demo to see the workflow in action.
How FluxForce strengthens Adverse Media Screening
FluxForce AI agents operate Adverse Media Screening in real time, capture audit-ready evidence automatically, and surface the gaps examiners cite before they become findings.