DSS: ניהול דינמי של ניתוח סמנטי למחיקת רעיונות בדגמי התפשטות

תקציר מקורי באנגליתarXiv:2604.16483v2 Announce Type: replace-cross Abstract: Text-to-image (T2I) diffusion models have introduced new security risks, as adversaries can exploit flexible text prompts to induce the generation of sensitive or policy-violating content (e.g., NSFW or copyrighted concepts). Concept erasure has emerged as a promising defense, aiming to suppress targeted semantics while preserving benign generation. However, existing approaches face a fundamental trade-off: training-based methods are costly and inflexible to emerging threats, while inference-time interventions often rely on unconstrained feature manipulation, leading to over-correction, semantic drift, and degraded utility. More critically, they lack an explicit understanding of local semantic structure, resulting in unstable behavi