Tell us about the role of your company in this project. 

IDIAP brings advanced research expertise in machine learning and natural language processing to TRACY. We focus on developing algorithms that can automatically identify patterns of criminal activity in non-content telecom data, ensuring that TRACY’s AI components are robust, scalable, and explainable.

Can you give an update about the latest developments? 

We have been working on optimising the filtering algorithms to handle NCD data and diverse telecom metadata formats. Our team has also been collaborating closely with the legal experts at Timelex to ensure the AI models adhere to ethical principles and privacy-by-design standards.

What are your company’s next steps in the project, in relation to the pilot site? 

Our next steps include refining the models based on pilot feedback, improving detection capabilities for specific crime scenarios, and supporting the integration of our algorithms into the TRACY system architecture.