SOURCE-LINKED INTELLIGENCE

Balancing Privacy, Utility, and Safety in LLM Alignment through Preference Optimization

arXiv · AI, language, vision and robotics · article · Aug 31, 2026 · UTC

Preference optimization is widely used to align large language models with human preferences, but preference-data composition may also influence privacy-relevant memorization. We examine whether adding synthetic privacy-preference pairs to Direct Preference Optimization (DPO) is associated with lower canary-based memorization signals without modifying the objective or introducing a formal privacy mechanism. We propose Privacy-Pressure Preference Mixing (P3M), a data-composition protocol that varies the amount of privacy-preference data while keeping helpfulness and harmlessness preference data

Read original source ↗ Open in workspace

recordType

paper

region

Global