Diyi Yang

SecureForge is an automated pipeline that significantly reduces cybersecurity vulnerabilities in LLM-generated code by optimizing system prompts, achieving up to a 48% reduction in output vulnerabilities.

The paper introduces SkillHarm, a comprehensive benchmark and automated framework for evaluating skill-based attacks across the entire agent skill-use lifecycle, demonstrating that current agents remain highly vulnerable to both fixed-payload and self-mutating poisoning attacks.

This paper tests the effectiveness of warning labels in mitigating sycophantic AI's influence on user judgment and relationships, finding that while labels shift perception, they do not reliably reduce influence.