MARS: A New Frontier in Multimodal LLM Safety
MARS, a training-free approach, enhances safety in multimodal LLMs by using textual refusal directions. Despite challenges, it promises consistent safety gains.
Read the full articleYou might also wanna read
Research Reveals LLM Refusal Behavior Is Controlled by a Single Direction in Model Activations
Conversational large language models are fine-tuned for both instruction-following and safety, resulting in models that obey benign requests
arXiv.org·2mo ago
BPE Tokenization Creates Exploitable Safety Gaps in LLM Alignment, Study Finds
Character-level perturbations bypass safety alignment in modern LLMs despite leaving prompts human-readable. We identify and test a central
arxiv.org·7d ago
The Ethical Dilemma of LLM Training Data and Content Creator Rights
An experimental strategy for contaminating Large Language Models
heydonworks.com·10mo ago
Synthesis of LLM Agent Failures: A Unified Taxonomy of Tool-Use, Planning, and Reasoning Limitations
Large language model (LLM) agents are increasingly evaluated on their ability to use tools, plan multi-step tasks, coordinate with other age
arxiv.org·2d ago
Cisco Researchers Find Multi-Turn Conversations Can Bypass LLM Safety Guardrails
Researchers at Cisco tested several well-known LLMs. They found of them could be tricked into bypassing guardrails, just through conversatio
infosecurity-magazine.com·1mo agoCisco Researchers Find Multi-Turn Conversations Can Bypass LLM Safety Guardrails
Researchers at Cisco tested several well-known LLMs. They found of them could be tricked into bypassing guardrails, just through conversatio
Infosecurity Magazine - Information Security & IT Security News and Resources·1mo ago
Stabilizing LLM Behavior: The Assistant Axis Approach to Preventing Harmful Persona Drift
Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.
anthropic.com·22d agoStabilizing LLM Behavior: The Assistant Axis Approach to Preventing Harmful Persona Drift
Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.
anthropic.com·22d ago
Comments
Sign in to join the conversation.
No comments yet. Be the first.