AML vendors deploy raw computing power to reduce false positives

Watch list filtering is every compliance officer's worst nightmare. With a single name like Muammar Gaddafi, spelled hundreds of different ways, and multiple watch lists to manage and update, the work is time consuming, costly and onerous.

Banks have often complained about the number of false positives generated, all of which need to be investigated. Some firms have even outsourced false positives' investigations to offshore locations in order to cope with the workload.

Anti-money laundering vendors are under pressure to not only reduce the number of false positives, but to make the filtering process more intuitive. In a report on Achieving Global Sanctions Compliance, Neil Katkov, senior vice president, Asia, Celent, says, "Achieving consistency in global sanctions compliance involves standardising operations, technology systems, and perhaps most essentially the compliance data—watchlists—that drive sanctions filtering.”

The lists themselves can be onerous and difficult to manage. Dr Tony Wicks, director, AML solutions, NICE Actimize, says HM Treasury's sanctions list in the UK had 3652 changed entries this year alone. There are also other lists banks need to comply with depending on the scope of their activity, including the well-known OFAC list, there is also an EU and UN watchlist and a Japanese FSA list.

The so-called Arab Spring has also had an impact on sanctions activity with new sanctions coming out associated with Iran. Anti-money laundering (AML) vendors like NICE Actimize say they are trying to lessen the workload for banks and the number of false positives using "fourth generation computational linguistics" - throwing raw computing power at "transliterate" words .

Explaining how the technology works, Wicks of NICE Actimize, says it can understand 800 million names, the context of those names and their cultural significance and make a "probalistic match" against the source of the name, which he says is important in terms of reducing false positives.