Veriteus provides a multimodal deception detection platform that analyzes speech, voice, micro-expressions, and behavioral patterns during digital form filling in real time. The system processes data without storing any personal information, ensuring full GDPR and EU AI Act compliance. It combines form, audio, and video modules that can be used independently or together, with internal benchmarks showing up to 40% reduction in fraud losses.
Funding
Funding not disclosed
Founders
Product
Problem
Traditional deception detection methods rely on voice stress analysis alone, which cannot distinguish natural stress or anxiety from actual deception and often requires storing sensitive voice recordings. These approaches produce high false positive rates and lack the transparency needed for audit-ready decision-making in fraud prevention, compliance, and investigative contexts.
Solution
Veriteus offers a multimodal behavioral AI platform that analyzes linguistic patterns, acoustic markers, micro-expressions, gaze patterns, and typing behavior to detect manipulation and inconsistency during digital interactions. The system operates in real time across three independent modules—form interface analysis, voice analysis, and video analysis—that can be deployed separately or combined for maximum accuracy. Every signal is cross-referenced and scored with clear confidence metrics, highlighting specific suspicious phrases, words, or expressions for human review. The platform processes data temporarily without storing any personal information, making it fully compliant with GDPR and EU AI Act high-risk classification requirements while producing audit-ready, explainable reports.
Target Audience
Primary customers are insurance companies detecting fraudulent claims before payout, corporate compliance teams conducting internal investigations, HR departments performing background verification, legal and forensic professionals gathering evidence, healthcare organizations preventing patient history fraud, and public sector agencies managing benefits fraud and regulatory compliance.
Features
- Form module detects hesitation, text edits, typing speed, answer order, and cognitive load patterns during digital form completion
- Audio module analyzes linguistic patterns, voice stress markers, and prosodic features, developed with Prof. Julia Hirschberg (Columbia University)
- Video module captures facial cues, gaze patterns, gesture inconsistencies, and physiological indicators via standard webcam
- Real-time analysis across all modules with active false positive prevention mechanisms
- Transparent confidence scores and explainable results that highlight specific phrases, words, and micro-expressions
- Zero PII architecture—no voice recordings or personal data stored, only temporary processing
- Modular design allowing independent or combined use of form, audio, and video analysis