Every data-protection program starts with the same unglamorous prerequisite: knowing what you have. You cannot apply the right control — encrypt it, restrict it, log access to it — to data you have not first identified as sensitive. At enterprise scale, across petabytes and millions of files, that classification problem is the real work.
SECURITI, Inc.'s US12216618B2, “System and a method for the classification of sensitive data elements in a file” (issued February 4, 2025; CPC G06F 21/6227 — protection of database/file data, and G06N 3/08 — neural-network learning methods), describes classifying sensitive data elements within a file using neural networks. Read it at US12216618B2.
Mechanically, the system uses neural networks to recognize sensitive elements — not just whole files, but the specific fields and values inside them — so governance can be precise rather than blunt. Element-level classification is what lets a policy distinguish a document that merely mentions a topic from one that actually contains regulated personal data, cutting the false positives that make coarse DLP unusable.
Why this is a business story: data security posture management (DSPM) became one of the hottest cloud-security segments of the mid-2020s, and accurate classification is its foundation — a DSPM product is only as good as its ability to find and label the sensitive data sprawled across cloud stores. The category drew significant acquisition activity as platforms moved to add DSPM, and the classification engine is the defensible core of any such product. Securiti positioned itself on exactly this data-intelligence-plus-governance thesis.
The grounded read: sensitive-data classification is the prerequisite for every data control, and doing it at the element level with neural networks is what makes governance precise. Securiti's 2025 grant names that classification engine — the foundation of the fast-growing DSPM category.