MAS-Shield: A Defense Framework for Secure and Efficient LLM MAS
Proposes a defense framework to secure and improve the efficiency of multi-agent LLM systems, addressing a key vulnerability in agentic AI deployments.
Proposes a defense framework to secure and improve the efficiency of multi-agent LLM systems, addressing a key vulnerability in agentic AI deployments.
Auto-scored: A real but partial technical mitigation step for securing multi-agent LLM systems, without evidence of deployment or verification.