Gradiant presents its advances in AI models for detecting terrorist threats, drug trafficking, and hate speech

The partners of PRESERVE, a European initiative led by Gradiant and launched in September 2024, are meeting this week in Bucharest to share progress, milestones, and define the next phases of the project.

The goal of PRESERVE is to detect and prevent cybercrime and terrorism while safeguarding privacy, ensuring fairness in law enforcement surveillance, and complying with the EU’s ethical standards for artificial intelligence.

Gradiant is participating this week in the third face-to-face meeting of the European alliance PRESERVE, held in Bucharest, to present the technological advances achieved in recent months. The initiative aims to detect and prevent cybercrime and terrorism, protect citizens’ privacy, ensure fairness in law enforcement monitoring, and meet EU ethical AI standards, while also improving police resource allocation.

Specifically, Gradiant has presented its progress in developing AI models supporting the project’s different use cases, as well as in defining and implementing a decentralised data capture, processing, and analysis architecture.

According to Luis Pérez Freire, Executive Director of Gradiant, “Currently, there is no platform like PRESERVE in the market. The innovation lies in bringing together privacy protection, ethical AI use, and improved decision-making in law enforcement procedures within a single platform”. He adds: “The project’s technological advantage comes from its ability to efficiently and securely collect and process large volumes of data from diverse sources”.

Main use cases of PRESERVE

The PRESERVE solution for privacy-preserving data analysis will be validated through four different use cases:

• Prevention and investigation of child sexual abuse, providing law enforcement authorities with advanced dashboards and visualisation tools to improve investigative processes, understand relationships between offenders and potential victims, explore their connections, and uncover hidden networks.

• Identification and mitigation of hate speech, through a platform offering interactive dashboards and reports to detect and address hate speech. The tool analyses trends, geographic hotspots, risk levels, and influential actors, correlating historical data with current incidents.

• Radicalisation risk, by providing advanced dashboards to enhance risk analysis and decision-making within Law Enforcement Authorities (LEAs). The system assesses the potential risk of radicalisation among users or communities and analyses its geographic prevalence.

• Drug trafficking detection, improving early detection by correlating police databases with online activity, including social networks and open sources. Using natural language processing, computer vision, and graph models, the platform identifies emerging substances, covert terminology, and organised networks. It also analyses public sentiment toward drug consumption to assess risk in specific areas, groups, or events, and optimises case analysis by applying spatial and temporal models to suspects, devices, and contacts.

During the meeting in Bucharest, representatives from the 13 consortium partners shared progress and milestones, refined end-user requirements, security measures, platform architecture, and functionalities, all aimed at preparing the toolkit for deployment by September 2027.

This project has received funding from the European Union’s Horizon Europe Research and Innovation programme under grant agreement N°101168309.