A Survey on Agentic Security: Applications, Threats and Defenses
Organizations: BRAC University · Qatar Computing Research Institute (QCRI)
Abstract
In this work we present the first holistic survey of the agentic security landscape, structuring the field around the fundamental pillars of Applications, Threats and Defenses. We provide a comprehensive taxonomy of 274 papers that covers red-teaming, blue-teaming and domain-specific applications, attack surfaces on agentic systems, defense techniques, and both offensive and defensive evaluation frameworks. For each pillar, we first discuss the literature around the taxonomy and then perform a quantitative analysis to understand the patterns across these works. For the applications, we show their security-lifecycle coverage map, compare between red-teaming and blue-teaming agents, and analyze the adversarial use of red-teaming applications. On the threat side, we analyze the entry points and agent-loop stages that attacks target, their specificity to the agentic setting, and the threat models they assume. On the defense side, we analyze the prevailing defense strategies, their cost and security trade-offs, and where in the agent lifecycle they are deployed. Finally, we provide a deatiled cross-cutting analysis, where we map which defenses cover which attack classes and chart trends in agent architecture, backbone model usage, and data modality coverage. Taken together, these findings give a researcher a structural understanding of how agentic capabilities, vulnerabilities, and countermeasures relate to one another, and to identify where the field is concentrated and where it is thin.
Figures & tables
| Abbr. | Meaning | Abbr. | Meaning |
|---|---|---|---|
| ASR | Attack Success Rate | SIEM | Security Information and Event Management |
| CVE | Common Vulnerabilities and Exposures | APT | Advanced Persistent Threat |
| CTF | Capture the Flag | IoC | Indicator of Compromise |
| SOC | Security Operations Center | MCP | Model Context Protocol |
| HITL | Human in the Loop | RAG | Retrieval-Augmented Generation |
| pass@ k | Success rate given k attempts |
| Defense Mechanism | Security Gain | Latency Cost | Financial Cost | Utility |
|---|---|---|---|---|
| PPA (Small Model) | High (Bypass Resistant) | 0.06 ms | No Overhead | Minimal |
| MCP-GUARD | +23.4% F1 Score | 455.9 ms | Low | High Utility |
| IPIGUARD | ASR 1% | 13.88 s | 14,605 Tokens per Task | 67.7% |
| PhishDebate | 94.14% Accuracy | 37.50 s | High | N/A |
| GuardAgent | 81.4% Accuracy | 6.30 s | 6,116 Tokens per Task | Moderate |
| PROVSEEK | 0.92 F1 Score | 30 min | 440K Tokens | N/A |
Appendix figures & tables2 assets
Supplementary material from the paper’s appendix.
Appendix
| Field group | Papers | # Fields | Example fields |
|---|---|---|---|
| Corpus metadata | 274 | 8 | paper identifier, pillars a paper is coded under |
| Structural | 274 | 9 | autonomy level, agent cardinality, architecture paradigm, cognitive roles, backbone models, knowledge sources, data modalities |
| Application | 113 | 3 | category, lifecycle stage(s), autonomy level |
| Threat | 86 | 15 | threat class, entry point, agent-loop stage targeted, attacker model, attacker goal, evaluation metric |
| Defense | 123 | 8 | trust boundary, cost/security tradeoff, defensive technique, defense positioning, defense coverage |
| Total | 43 |
AI Agent Security
Artificial intelligence agent security studies attacks and safeguards for tool-using, persistent, and collaborative systems across digital and physical settings.
See all 553 papers