Privacy-Enhancing Technologies for Visual Data in AI

Visual data is central to a growing range of AI-driven products and services in autonomous vehicles, robotics, wearables, smart environments, and public and private spaces. It is among the most challenging categories of personal data to process lawfully under the GDPR and the AI Act. Controllers, processors and deployers face a consistent set of questions around lawful basis, data minimization, transparency, data subject rights, transfer analysis and security, all while supporting ambitious innovation roadmaps. Privacy-enhancing technologies, from anonymization and pseudonymization to synthetic data, on-device processing and federated learning, offer a growing toolkit for addressing these challenges. However, their application to images and video raises practical and technical questions that cut across legal and engineering disciplines. This panel convenes legal, compliance, regulatory and engineering practitioners to share their practical experience of enabling human-centric AI innovation. Discussion will touch on practical experience from regulatory sandbox projects, including lessons learned from working with real-world use cases, supervisory engagement and the realities of selecting and deploying PETs in production AI systems. The session is intended as an accessible, collaborative exchange.

Sponsored by Syntonym