New Patents Protect Company’s Industry Lead in Recognizing and Semantically Grouping and Categorizing Data Records, and Contextualizing Anomalous Behavior Towards Data
SAN JOSE, Calif., July 16, 2025–(BUSINESS WIRE)–Concentric AI announced today that it has secured its fifth and sixth patents of the year, emphasizing the company’s ongoing innovation and leadership in advancing AI-driven data security. These most recent patents reinforce the company’s industry leadership in recognizing, semantically grouping, and categorizing data records, as well as contextualizing anomalous behavior specifically related to data. These unique capabilities set Concentric AI apart from all other data security tools on the market.
The first of the two newly granted patents, “Method and Electronic Device to Assign Appropriate Semantic Categories to Documents with Arbitrary Granularity,” recognizes the company’s approach to harnessing Large Language Models (LLMs) for semantic categorization. This method assigns semantic categories and subcategories to content with unmatched accuracy at enterprise scale. While LLMs are very data-intensive, highly optimized data compression techniques called adaptive manifold compression models achieve up to 10 times more compression and significantly improve data discovery rates, leading to better compute efficiency, storage, and performance.
A major benefit of the technology described in the new patent is that there’s no requirement to specify the entire taxonomy of categories upfront; the design allows for dynamically adding or removing categories seamlessly, and categories can differ in levels of detail. For example, they could be as specific as certain types of tax forms or as broad as all legal agreements. This feature is foundational to Concentric AI’s product and enables companies to categorize data across repositories into meaningful semantic groups without any upfront effort, such as configuring regexes and patterns.
The second patent, “Methods and Systems for Determining Anomalous User Access to Data Objects,” underscores the company’s novel approach to legacy user behavior analytics (UBA). Traditional UBA and UEBA (user and entity behavior analytics) tools aim to detect insider threats by profiling user activity and spotting anomalies but often generate false positives because many “anomalies” are benign behaviors with valid business reasons. These conventional methods fall short because they lack the context of the data associated with user behavior. With this new patent, Concentric AI integrates data context with user behavior, enabling the accurate identification and pinpointing of risky behaviors related to sensitive data with unmatched precision.
