Alignment Drift in Multimodal LLMs: A Two-Phase, Longitudinal Evaluation of Harm Across Eight Model Releases

Documenti analoghi

• Red Teaming Multimodal Language Models: Evaluating Harm Across Prompt Modalities and Models
  di: Van Doren, Madison, et al.
  Pubblicazione: (2025)
• Same Model, Different Weakness: How Language and Modality Reshape the Jailbreak Attack Surface in Frontier MLLMs
  di: Ford, Casey, et al.
  Pubblicazione: (2026)
• CUPID: Evaluating Personalized and Contextualized Alignment of LLMs from Interactions
  di: Kim, Tae Soo, et al.
  Pubblicazione: (2025)
• Multimodal Transformer Models for Turn-taking Prediction: Effects on Conversational Dynamics of Human-Agent Interaction during Cooperative Gameplay
  di: Bae, Young-Ho, et al.
  Pubblicazione: (2025)
• Mitigating Harmful Erraticism in LLMs Through Dialectical Behavior Therapy Based De-Escalation Strategies
  di: Rangarajan, Pooja, et al.
  Pubblicazione: (2025)