Tag: LLM Vulnerabilities
Browse our exclusive articles!
How Large Language Models Generate Harmful Content
Discover how large language models produce harmful content via a unified mechanism and explore new strategies to improve AI safety and alignment.
LogicPoison: Logical Attacks on GraphRAG Systems
Discover how LogicPoison disrupts Graph Retrieval-Augmented Generation by targeting logical graph structures, bypassing defenses and degrading AI reasoning...
Environment-Injected Memory Poisoning Attacks on Web Agents
Discover how environment-injected memory poisoning attacks exploit web agents across sites and sessions, highlighting urgent AI browser security risks.
Finetuning Triggers Verbatim Recall in Large Language Models
Study reveals finetuning LLMs can cause verbatim recall of copyrighted books, exposing major copyright and safety vulnerabilities in AI models.
Stealthy Injection Attacks on Model Context Protocols
Discover how tree-based adaptive search generates stealthy injection payloads to exploit Model Context Protocol vulnerabilities in AI systems.