Position Paper: Lifelogging Video Streams Make the Privacy-Utility Trade-off Inevitable

A position paper argues that the privacy-utility trade-off of lifelogging video streams is an inevitable foundational challenge for next-generation AI systems. Although video streams can improve the utility of proactive agents and world models, they expose sensitive information such as behavioral patterns, emotional states and social interactions. Existing protections either target only specific attacks or incur excessive utility loss, and do not cover the full data utilization pipeline; the paper calls for joint optimization of utility and privacy across the full pipeline, and for developing formal leakage metrics and standardized benchmarks.