I-HITL: Inalienable Human-in-the-Loop
Introduces I-HITL (Inalienable Human-in-the-Loop), a security protocol ensuring AI agents require explicit human approval before taking actions, even if the system is compromised by the agent.
Read paperAI & multi-agent security
Research library.
Explore the research. Find your next question.
Browse papers by security topic, read concise summaries, and follow the original sources.
Recent publications
Recent papers
The Agent in the Middle: How AI-Mediated Access Reshapes Web Vulnerability Exploitation
Web services increasingly expose backend interfaces for AI agents to perform tasks directly, altering traditional web vulnerability exploitation methods.
Read paperThen there is prompt injection, a malicious email, ...: The Wild Wide West of OpenClaw Agentic System Security
OpenClaw is an AI-powered personal assistant framework combining LLM reasoning, persistent memory, tool invocation, and workflows, enabling automation but introducing complex multi-stage security risks due to lack of explicit trust boundaries.
Read paperResearch trends
Papers by topic
4545 papers
3079 papers
1418 papers