Time-series data is increasingly used in sensitive applications such as energy monitoring, healthcare, and finance, raising critical challenges at the intersection of utility, privacy, and trust. In this talk, I will present perspectives on these challenges through three complementary research directions. First, I will introduce TELESAFE, an unsupervised and non-intrusive system that leverages time-series patterns to detect work–private boundary crossings from energy consumption data, achieving performance comparable to supervised approaches without requiring labeled data. Second, I will revisit the common assumption that transformed representations, such as the Matrix Profile, provide inherent privacy protection. I will show that these representations remain vulnerable to privacy attacks, including singling-out, linkability, and inference, and can even enable reconstruction of sensitive time series patterns. Finally, I will explore how to restore trust in time-series analytics. For that, I will present two complementary approaches: (i) verifiable analytics using Zero-Knowledge Proofs, enabling users to prove properties over their data without revealing it, and (ii) an toolbox for selecting Local Differential Privacy mechanisms and privacy budget epsilon, under utility–attackability trade-offs.