Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM

Documenti analoghi

• UCD: Unlearning in LLMs via Contrastive Decoding
  di: Suriyakumar, Vinith M., et al.
  Pubblicazione: (2025)
• Algorithmic Pluralism: A Structural Approach To Equal Opportunity
  di: Jain, Shomik, et al.
  Pubblicazione: (2023)
• When Style Breaks Safety: Defending LLMs Against Superficial Style Alignment
  di: Xiao, Yuxin, et al.
  Pubblicazione: (2025)
• Unstable Unlearning: The Hidden Risk of Concept Resurgence in Diffusion Models
  di: Suriyakumar, Vinith M., et al.
  Pubblicazione: (2024)
• AI Generated Child Sexual Abuse Material -- What's the Harm?
  di: Ciardha, Caoilte Ó, et al.
  Pubblicazione: (2025)