[{"data":1,"prerenderedAt":4},["ShallowReactive",2],{"article-body-95-percent-false-positives-is-a-design-choice":3},"\nSanctions screening tools generate vast numbers of alerts, and\ncompliance teams clear them every day. False-positive rates above\n95% are often accepted as an unavoidable part of managing risk.\n\nIt feels reasonable. Sanctions enforcement is a difficult problem.\nHundreds of thousands of sanctioned individuals and entities sit\nacross multiple lists, names change between languages, and aliases\nare often designed to obscure who is really involved.\n\nBut the complexity of sanctions is not what creates most of the\nnoise. Most screening systems check names. The law is\nconcerned with ownership. So analysts spend hours proving that\nthis Rodriguez is not that Rodriguez, while the real risk may sit\ninside a perfectly ordinary company name that appears on no\nsanctions list at all.\n\nThe tools check the name on the door. The law cares who owns the\nbuilding.\n\n## What the law actually says\n\nOFAC, the US sanctions authority, makes the distinction explicit.\nIts 50 percent rule states that a company owned half or more,\ndirectly or indirectly, by one or more sanctioned parties is\nitself blocked, even if its own name appears on no list. And\nownership is aggregated. If two sanctioned parties own 30% and 25%\nof the same company, their interests combine, and the company is\nblocked even though neither shareholder individually holds a\ncontrolling stake. US export controls have been moving the same\nway.\n\nThe implication is straightforward. The critical compliance\nquestion is not whether a counterparty's name resembles one on a\nwatchlist. It is who ultimately owns the counterparty, and how\nmuch sanctioned ownership exists throughout its structure.\n\n## Why name screening fails twice\n\nMeasured against that standard, name-based screening breaks down\nin two different ways.\n\nIt generates alerts where the law sees little or no risk. Every\ncompany whose name resembles a listed entity becomes a potential\nmatch, and in the overwhelming majority of cases the investigation\nfinds no connection to any sanctioned party. That is where the 95%\ncomes from.\n\nAnd it misses the situations the law is explicitly designed to\ncatch. A trading company owned through multiple layers of holding\ncompanies may have a completely clean name. No match appears, no\nalert is raised, and the entity may still be blocked because of\nwho ultimately owns it. The system stays silent precisely where\nthe regulation applies.\n\nBoth failures stem from the same design choice, which is screening\nnames instead of screening ownership. That is why reducing false\npositives and improving detection are not separate problems. They\nhave the same solution.\n\n## The cost of getting it wrong\n\nEvery false positive consumes analyst time. Every false negative\ncreates exposure, and the exposure now runs to nine figures. BIS,\nthe US export-control enforcement agency, settled with\n[Applied Materials for $252.5 million in February 2026](https://www.bis.gov/press-release/applied-materials-pay-252-million-penalty-bis-illegally-exporting-semiconductor-manufacturing-equipment),\nthe second-largest civil penalty in its history behind Seagate's\n$300 million in 2023. Regulators increasingly expect organisations\nto understand who stands behind their counterparties, not simply\nwhat they are called.\n\n## Screen what the law screens\n\nStart by working out which real-world company each counterparty\nactually is, so that multiple spellings, aliases and local\nregistrations stop generating separate investigations. Then build\nthe ownership chain, tracing owners through each corporate layer\nand adding up sanctioned ownership across the structure. Then\napply the regulatory test continuously, so a change in a parent\ncompany's ownership or status automatically updates the assessment\nof every subsidiary beneath it.\n\nThe output changes completely. Instead of \"this name resembles a\nlisted name\", the analyst sees \"this counterparty is 55% owned,\nthrough these ownership chains, by these sanctioned parties, as of\nthis date\", with the evidence attached. The queue shrinks, the\nquality of what remains rises, and analyst time moves from\nuntangling names to making compliance decisions.\n\n## Measure what matters\n\nThe true measure of a screening programme is not how many alerts\nit generates. It is whether the analyst's next hour goes on\nassessing genuine risk or on untangling spelling variations.\n\n[The Prophesee Compliance Suite](/solutions/compliance/trade)\nscreens ownership structures, applies ownership rules continuously\nand updates assessments as corporate structures change.\n\nBecause the law does not screen names. It screens ownership. Your\nsystem should too.\n[Run your counterparty file through it](/contact).\n",1786984935415]