Jul 2026
When T2I Synthetic Data Backfires: Amplified Privacy Risks in Real-Synthetic Mix Training
This work reveals, for the first time, that RSMT can substantially amplify privacy leakage of these real training samples, and proposes a lightweight leakage propensity indicator computable from real data alone that reliably identifies high-risk datasets unsuitable for entering RSMT, as a self-assessable mitigation.
Na Li, Boyu Kuang, Hongsheng Hu et al.
· arXiv.org · 0 citations