How to Run an Adverse Media Search

• FATF. Its risk-based guidance for banks lists "verifiable adverse media searches" among examples of enhanced due diligence, and its PEP guidance notes that internet and media searches are widely used but may not always be comprehensive or reliable.
• UK. The JMLSG guidance asks firms whether there are adverse media reports about the customer, to judge credibility on the quality and independence of the source and the persistence of reporting, and notes that the absence of a conviction alone may not be enough to dismiss allegations. Adverse media searches are listed among enhanced due diligence measures.
• Singapore. MAS's Guidelines to Notice 626 expect screening policies to set out the sources used, including commercial databases and, where appropriate, pertinent search engines — including engines used mainly in the person's country of nationality, residence or source of wealth. Using public sources to understand a customer's reputation is listed as an enhanced measure, and adverse news about financial crime may warrant a suspicious transaction report.
• United States. The FFIEC examination manual expects banks to decide, on a risk basis, when to obtain more information on higher-risk customers, for example through negative media search programmes, and lists their results as a reason to review a relationship.

Wolfsberg recommends defining search strings that combine the name with risk keywords, using exact matching for names and multi-word phrases, and keeping search settings consistent. There is no universal string; build your own keyword set around the risks that matter to you, such as:

• Financial crime: fraud, money laundering, bribery, corruption, embezzlement, tax evasion, sanctions, terrorism, trafficking.
• Legal process: arrested, charged, indicted, convicted, sentenced, investigation, fined, penalty, regulator, lawsuit.
• Local languages: the same terms in the languages of the subject's nationality, residence and business.

Google's help page documents quotes for exact phrases, the minus sign, site:, before: and after:, and filetype:. Other operators analysts often use, such as OR, are not in Google's current documentation, and some older ones — including cache: and related: — no longer work, so test strings before relying on them and always record the exact string used.

• Confirm it is the same person or company. Compare the identifiers in the article with what you hold: age or date of birth, location, role, employer, associates. Wolfsberg gives the example of allowing a date-of-birth tolerance of one year either side when filtering alerts.
• Judge credibility. Consider the type of outlet, editorial oversight, the wider context in the country concerned, whether other independent sources corroborate the story, and the quality of the website. Be alert to disinformation.
• Judge relevance. Is the story about financial crime or conduct that matters to your risk assessment, how recent is it, and what was the outcome — allegation, charge, settlement, acquittal or conviction?
• Do not dismiss on technicalities. An allegation that was never prosecuted can still be relevant; the JMLSG guidance notes that the absence of criminal convictions alone may not be sufficient to dismiss allegations.

Many teams try to script Google searches — the question "how do I run an adverse media check through Google search in C#?" comes up often. Before you build anything, note:

• Google's Custom Search JSON API is closed to new customers, and existing customers have until 1 January 2027 to move to an alternative. From that date, new Programmable Search Engines can search only specified sites (up to 50 domains), not the whole web.
• Scraping results is not permitted. Google's terms prohibit automated access to its services in breach of the machine-readable instructions on its pages, and its robots.txt disallows automated access to search results.
• Search engines were not designed to assess financial crime risk and may push the most relevant results down the page, as Wolfsberg notes.

For scale, firms generally use licensed news data or structured adverse-media datasets that are already categorised by risk type, and keep manual searching for enhanced due diligence and to resolve possible matches.