Research Dashboard

AI Security Research Portal

AI security second brain overview: source evidence, priority research questions, concepts, methods, and the topic graph connecting them.

Research Questions Current Synthesis Browse Wiki

A lightweight map of the main topic clusters. Use the concept and research question pages for deeper reading.

260Sources

11Research Questions

13Concepts

6Methods

Recent Sources

Priority Research

Research Questions Indexindex RQ-20260702-001-agent-securityresearch-question RQ-20260702-002-benchmark-validityresearch-question RQ-20260702-003-defense-generalizationresearch-question RQ-20260702-004-agent-protocol-securityresearch-question RQ-20260702-005-memory-poisoning-defenseresearch-question

Topic Clusters

Agent Identity and Authorizationconcept Agent Security and Tool Abuseconcept AI Cybersecurity Operationsconcept AI Security Governance and Standardsconcept AI Security Taxonomyconcept Concepts Indexindex Evaluation Benchmarks for AI Securityconcept MCP and Agent Protocol Securityconcept Memory Poisoning and Agent Stateconcept

Start Here

Current Synthesissynthesis indexindex Raw Corpus Synthesis 2026-07-02synthesis Research Questions Indexindex

Methods

Benchmark-Based Security Evaluationmethod Evidence Grading for AI Securitymethod Methods Indexindex Red Teaming Agentic AImethod Runtime Monitoring and Agent Gatewaysmethod Threat Modeling Agentic Systemsmethod

Raw source catalog preview

\"What Happens Locally, Leaks Globally\": Detecting Privacy Leakage Risks in MCP ServersSources A Layered Security Framework Against Prompt Injection in RAG-Based ChatbotsSources A Practical Guide for Securely Using Third-Party MCP ServersSources A Survey of Self-Evolving Agents: What, When, How, and Where to Evolve on the Path to Artificial Super Intelligenceraw_capture A Survey on Model Extraction Attacks and Defenses for Large Language ModelsSources A user-centric machine learning framework for cyber security operations centerSources Adversarial Machine Learning: A Taxonomy and Terminology of Attacks and MitigationsSources AECR: Automatic attack technique intelligence extraction based on fine-tuned large language modelSources Agent Security Bench (ASB): Formalizing and Benchmarking Attacks and Defenses in LLM AgentsSources Agent Security Bench (ASB): Formalizing and Benchmarking Attacks and Defenses in LLM-based Agentsraw_capture Agent-Assisted Side-Channel Attacks on Non-Prefix KV Cache in RAGSources AgentCanary: A Security Evaluation Framework for Autonomous AI Agents in Real Executable EnvironmentsSources AgentCyberRange: Benchmarking Frontier AI Systems in Realistic Cyber RangesSources AgentDojo: A Dynamic Environment to Evaluate Prompt Injection Attacks and Defenses for LLM Agentsraw_capture AgentDyn: Are Your Agent Security Defenses Deployable in Real-World Dynamic Environments?Sources AgentEvolver: Towards Efficient Self-Evolving Agent SystemSources Agentic AI Security: Threats, Defenses, Evaluation, and Open ChallengesSources AgenticCyOps: Securing Multi-Agentic AI Integration in Enterprise Cyber Operationsraw_capture AgenticOS: An Intent-Oriented Secure Operating System Architecture for Autonomous AI AgentsSources AgentLens: Interpretable Safety Steering via Mechanistic Subspaces for Multi-Turn Coding AgentSources AgentOps: Enabling Observability of LLM AgentsSources AgentPoison: Red-teaming LLM Agents via Poisoning Memory or Knowledge BasesSources AgentRiskBOM: A Risk-Scoping Security Bill of Materials for Agentic AI SystemsSources Agents All the Way Down; A Methodology for Building Custom AI Agents from Substrate to ProductionSources