TENSOR at a glance: Improving European investigation capabilities to fight terrorism and serious crime by leveraging the latest advancements in modern biometric technologies

Co-funded by the European Union under Grant Agreement no. 101073920. Views and opinions expressed are however those of the authors only and do not necessarily reflect those of the European Union or the European Research Executive Agency. Page 1 of 6 TENSOR at a glance: Improving European investigation capabilities to fight terrorism and serious crime by leveraging the latest advancements in modern biometric technologies The TENSOR project is set to enhance public safety in the EU by revolutionising the investigation capabilities of forensic institutes and Police Authorities through the use of emerging biometric technologies. This 3-year innovation action, supported by the Horizon Europe Programme, has brought together 20 partners from 15 European countries on the way to accelerate the adoption of modern biometric solutions in law enforcement and create a common ground among security practitioners and researchers by introducing the first European Biometric Data Space. Challenge The core challenge addressed by the TENSOR project lies in the effective integration of modern biometric technologies into forensic processes to enhance suspect identification in line with the evolving operational demands of law enforcement. Although biometric identification technologies have advanced significantly in recent years – largely through the application of AI – Law Enforcement Agencies (LEAs) continue to face practical barriers to their seamless adoption. TENSOR tackles this issue by delivering advanced suspect identification capabilities through an innovative multi-modal biometric fusion approach. This includes facial, voice, fingerprint, and gait recognition, along with behavioural patterns derived from mobile device usage. These diverse biometric modalities are intelligently combined and interpreted to reflect the real-world investigative contexts and time-sensitive scenarios in which LEAs operate. More specifically, TENSOR aims to equip Police Authorities with advanced, explainable, and actionable insights, enabling them to make lawful and well-informed decisions. This responds to the growing demand for enhanced forensic capabilities and lawful evidence collection, supporting the apprehension of criminals and the presentation of robust biometric evidence in court. By emphasizing explainable AI and human-in-the-loop mechanisms, TENSOR promotes Project Overview Co-funded by the European Union under Grant Agreement no. 101073920. Views and opinions expressed are however those of the authors only and do not necessarily reflect those of the European Union or the European Research Executive Agency. Page 2 of 6 transparency and accountability in biometric analysis, avoiding the risks of black-box decisionmaking while remaining aligned with European legal and ethical standards. Finally, TENSOR further contributes to the prevention, detection, and deterrence of various forms of crime by facilitating suspect identification through the analysis of behavioural data linked to the owners of mobile devices associated with criminal activity. The platform’s ability to integrate fragmented and heterogeneous data sources enhances situational awareness for LEAs, enabling them to more effectively combat organized crime. By doing so, TENSOR contributes to the security of EU citizens through timely, reliable, and precise identification of suspects. The TENSOR Solutions The TENSOR project aims to revolutionize the way LEAs across Europe identify suspects and combat serious crime and terrorism. In an era of increasingly sophisticated threats and fragmented investigative workflows, TENSOR provides innovative, AI-powered solutions for the extraction, analysis, storage, and secure cross-border sharing of biometric evidence. The platform supports automated, robust, privacy-preserving, and scalable operations that respect EU data protection regulations while enabling seamless international cooperation. Figure 1. TENSOR's technical solutions • Fusion of Physiological and Behavioural Biometrics: Advanced multi-modal biometric identification system that supports law enforcement agencies in identifying suspects for combating crime and terrorism more effectively. The platform integrates a broad spectrum of biometric modalities, combining well-established methods, such as fingerprints, facial and voice recognition, with emerging behavioural biometrics, including gait recognition (based on unique walking patterns) and keystroke dynamics (based on typing behaviour such as timing, rhythm, and pressure) by dynamically combining complementary biometric evidence using TOPSIS, a quantitative and mathematical multicriteria decision analysis (MCDA) method. • Intelligence from Soft Biometrics: Derives a rich set of soft biometric traits, such as age, gender, body proportions, and language use, from multiple modalities to enrich suspect profiling and support cross-checking when hard biometric matches are unavailable. Project Overview Co-funded by the European Union under Grant Agreement no. 101073920. Views and opinions expressed are however those of the authors only and do not necessarily reflect those of the European Union or the European Research Executive Agency. Page 3 of 6 • Explainability for Fair and Unbiased Biometric Technologies: Embedded explainability mechanisms and expert-in-the-loop workflows, enhancing the system’s decisionmaking pipeline. • Biometrics Dataspace: Enables suspect identification in transnational cases by performing secure biometric data matching (e.g., facial images, voiceprints, fingerprints) across jurisdictions without exposing sensitive information. Built on IDSA data space technology, TENSOR uses privacy-enhancing technologies like homomorphic encryption to preserve privacy during processing, and blockchain-based smart contracts for automated access control and lawful data use. • Digital Forensics: Develops methodologies for extracting data of interest (e.g., app usage patterns, location traces, configuration data from health apps, media files, text messages, etc.) from seized mobile devices, bypassing the inherent encryption mechanisms. • AI-powered Digital Assistant: Features an AI-powered Digital Assistant to support investigators, which consolidates biometric identification results into structured, natural language reports. Through a chat-based interface, the assistant explains how TENSOR's technologies work, helps users navigate the platform, and provides step-by-step guidance for key investigative tasks. The TENSOR Pilot Use Cases TENSOR is built around three overarching goals, each validated through a real-world pilot use case. These goals reflect TENSOR’s commitment to improving biometric identification capabilities, digital forensic investigations, and secure data exchange across national borders. Pilot Use Case 1: Evidence collection through intelligence derived from correlated physiological and behavioural biometrics based on CCTV footage. Objective: Fuse multiple biometric traits for more reliable and accurate suspect identification. Description: TENSOR’s first objective is to deliver an advanced multi-modal biometric identification system that strengthens the ability of law enforcement agencies to identify suspects for combating crime and terrorism more effectively, streamline fragmented workflows, and improve investigative efficiency. The platform integrates a broad spectrum of biometric modalities, combining well-established methods, such as fingerprints and facial recognition, with emerging behavioural biometrics, including gait recognition (based on unique walking patterns) and keystroke dynamics (based on typing behaviour such as timing, rhythm, and pressure). To support investigators, TENSOR also features an AI-powered Digital Assistant, which consolidates biometric results into structured, natural language reports. Through a chatbased interface, the assistant explains how TENSOR's technologies work, helps users navigate the platform, and provides step-by-step guidance for key investigative tasks. Project Overview Co-funded by the European Union under Grant Agreement no. 101073920. Views and opinions expressed are however those of the authors only and do not necessarily reflect those of the European Union or the European Research Executive Agency. Page 4 of 6 Pilot Use Case 2: Digital forensics extensions allowing orphan device owner identification. Objective: Identify the owner of an orphan mobile device through the extracted behavioural data. Description: The second objective addresses the growing need for advanced digital forensics tools capable of unlocking intelligence from orphan mobile devices, such as smartphones, tablets, or laptops seized during criminal investigations that lack immediate ownership information. As organized crime increasingly operates in digital spaces, mobile devices often contain critical data such as text inputs, GPS records, photos, transaction logs, and other behavioural traces that can establish connections, reconstruct timelines, and identify those involved. TENSOR applies digital forensics methods to extract and analyse these traces, helping investigators infer attributes of the user such as past locations, communication patterns, language proficiency, and even education level. This process transforms raw device data into actionable insights that support law enforcement in linking suspects to crimes and uncovering hidden networks. Recognizing that many LEAs lack the budget and specialized personnel for acquiring advanced tools, TENSOR’s solution is designed to be accessible and cost-effective, enabling even non-specialized LEAs to overcome the limitations of existing commercial tools. Project Overview Co-funded by the European Union under Grant Agreement no. 101073920. Views and opinions expressed are however those of the authors only and do not necessarily reflect those of the European Union or the European Research Executive Agency. Page 5 of 6 Pilot Use Case 3: Biometric data protection and secure exchange in a cross-border scenario. Objective: Implement the first Biometrics Data Space ecosystem for sovereign and trustworthy cross-border data exchange. Description: The third objective is the creation of a Biometrics Data Space, a secure, privacypreserving digital ecosystem that facilitates cross-border biometric data sharing among LEAs and forensic agencies. The Biometrics Data Space enables the identification of suspects in transnational cases by allowing agencies to securely search for and match biometric data (e.g., facial images, voiceprints, fingerprints) held in other jurisdictions, without disclosing sensitive raw personal data. Built on the data space technology, the TENSOR framework integrates Privacy Enhancing Technologies (PETs), such as homomorphic encryption, to ensure that data privacy is preserved even during processing. It also employs blockchain-based smart contracts to enforce automated access control and govern the lawful use of biometric data. These mechanisms ensure that data access is limited to authorized entities and that usage policies defined by data owners are strictly respected. By fostering sovereign, trusted, and legally compliant biometric data exchange, TENSOR aims at stronger international collaboration, faster suspect identification, and more effective resolution of cross-border investigations. Project Overview Co-funded by the European Union under Grant Agreement no. 101073920. Views and opinions expressed are however those of the authors only and do not necessarily reflect those of the European Union or the European Research Executive Agency. Page 6 of 6 TENSOR Fact Sheet Topic • HORIZON-CL3-2021-FCT-01-05 - Modern biometrics used in forensic science and by police Dates • Start Date: January 1st, 2023 • End date: December 31st, 2025 • Duration: 36 months Consortium 20 partners from 15 countries • 6 x RTOs • 4 x Police Authorities & forensic institutes • 3 x large industries • 7 x SMEs Pilots • Evidence collection through intelligence derived from correlated physiological and behavioural biometrics based on CCTV footage @ Police of the Czech Republic • Digital forensics extensions allowing orphan device owner identification @ Ministry of the Interior, Finland • Biometric data protection and secure exchange in a cross-border scenario @ Criminal Investigation Police, Portugal @ General Police Inspectorate, Moldova Funding • TENSOR has received € 4 562 975 from European Union’s Horizon Europe research and innovation programme under grant agreement No 101073920 (Total cost: € 5 739 725). TENSOR Partners Contact Us tensor-horizon.eu @ [email protected] TENSOR TENSOR TENSOR World with solid fill