Behavox, an AI company specializing in transforming data into insights for business safeguarding and enhancement, has announced the general availability of its proprietary GPU-powered AI Risk Policies (AIRPs). This release is designed to strengthen Behavox’s capabilities in delivering value and efficiency for its customers.
These GPU-powered AIRPs build upon the success of Behavox’s existing CPU-based AIRPs, which have demonstrated a reduction of alerts by over 80% compared to traditional lexicon-driven approaches. With the new GPU-powered iteration, customers can expect an additional average 60% reduction in alerts beyond this baseline, alongside an improvement in true positive detection rates to over 90%. These performance enhancements are provided at no additional cost to customers, aligning with Behavox’s goal of lowering the total cost of ownership for compliance programs.
Behavox utilizes a proprietary Large Language Model (LLM), offering customers full transparency, explainability, and control, a differentiator from vendors reliant on open-source solutions. The models have undergone rigorous model risk and governance testing and have been included in customer filings to regulators. This vetting by governance teams, monitors, consultants, and regulatory bodies underscores their suitability for highly regulated markets.
Behavox’s GPU models maintain consistently high recall across more than 15 languages, accommodating regional and accent variations. This enables global compliance teams to operate with confidence. For instance, average recall rates include English at 89.71%, Spanish at 87.61%, Japanese at 93.24%, French at 93.81%, Danish at 94.52%, German at 93.22%, Swedish at 94.89%, Italian at 93.56%, Portuguese at 94.10%, Chinese at 94.66%, Korean at 95.20%, Hindi at 93.89%, Indonesian at 94.37%, Vietnamese at 93.41%, and Tagalog at 94.37%.
The GPU infrastructure also paves the way for future developments, including expansion to over 50 languages in upcoming releases and the groundwork for holistic surveillance, such as integrated trade surveillance. The GPU-powered AIRPs are available to all customers today and support self-service installation and validation.
Manish Kumar, Head of Product Management at Behavox, stated, “Compliance teams face mounting budgetary pressures while striving for efficiency. Our GPU-powered models significantly reduce alert fatigue while raising detection quality — delivering more value without additional cost.”
Organizations interested in independently validating the performance of GPU-backed AI Risk Policies can request a free trial of Behavox Pathfinder.
Founded in 2014, Behavox is an AI company headquartered in London, with additional offices in New York City, Montreal, Seattle, Singapore, and Tokyo. The company’s proprietary technology and industry-specific LLM enable users to analyze structured and unstructured corporate data to detect compliance risks, streamline data archiving, predict regulatory breaches, and derive insights from big data. Behavox serves a global client base that includes banks, hedge funds, commodities firms, private equity firms, crypto firms, and other non-financial companies.
