CoChat

Agentic AI Security: Threats, Defenses, Evaluation, and Open Challenges

Open access

The DOI matches its Crossref record; 264 of the 288 references it cites were found in a registry or an index.

Abstract

Agentic AI systems powered by large language models (LLMs) and endowed with planning, tool use, memory, and autonomy, are emerging as powerful, flexible platforms for automation. Their ability to autonomously execute tasks across web, software, and physical environments creates new and amplified security risks, distinct from both traditional AI safety and conventional software security. This survey outlines a taxonomy of threats specific to agentic AI, reviews recent benchmarks and evaluation methodologies, and discusses defense strategies from both technical and governance perspectives.We synthesize current research and highlight open challenges, aiming to support the development of secure-by-design agent systems.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1109/access.2026.3675554

Cites

264 of 264 references verified0 citation errors · 0 not checkable

Statistics

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Items 201–264 of 264

Research works printed but not linked: 70

  1. 10.1007/978-981-97-5656-8 · DOI
  2. Work, 2024, AI what do large language models ‘understand, Image, 21, 1
  3. 10.1007/978-3-031-54827-7_5 · DOI
  4. 2025, AI Agents Are Here. So Are the Threats.
  5. 2024, LangChain Documentation
  6. Richards, 2024, AutoGPT: An Autonomous GPT Experiment
  7. Dal Cin, 2025, Three essentials for agentic AI security, MIT Sloan Manage. Rev., 1
  8. 2025, Just in time? Manufacturers Turn to AI to Weather Tariff Storm
  9. 2025, Forget Chatbots. AI Agents are the Future
  10. 10.1016/j.cell.2024.09.022 · DOI · Paper page
  11. 2025, Inside the Automated Warehouse Where Robots Are Packing Your Groceries
  12. 2025, Amazon’s Delivery, Logistics Get AI Boost
  13. 10.1007/978-3-031-90026-6_10 · DOI · Paper page
  14. Micro, 2025, Preventing Zero-Click AI Threats: Insights From EchoLeak
  15. 2025, AI: Advent of Agents Opens New Possibilities for Attackers
  16. 10.52202/079017-4136 · doi:10.52202/079017-4136 · Paper page
  17. 10.18653/v1/2025.emnlp-main.1771 · DOI
  18. 2025, When Llms Autonomously Attack
  19. 2025, OWASP GenAI LLM01: Prompt Injection
  20. 10.18653/v1/2025.acl-long.890 · DOI
  21. 10.18653/v1/2023.emnlp-main.849 · DOI
  22. Park, 2025, Unveiling AI Agent Vulnerabilities Part II: Code Execution
  23. 2024, CVE-2024-5565: Vanna.AI Remote Code Execution Vulnerability
  24. Chhabra, Towards fair video summarization
  25. 10.1609/aaai.v34i07.6918 · DOI
  26. Rehberger, 2024, DeepSeek AI: From Prompt Injection To Account Takeover
  27. 2025, Technical Blog: Strengthening AI Agent Hijacking Evaluations
  28. Kennedy, 2011, Metasploit: Penetration Tester’s Guide
  29. Muehlberger, 2020, CSRF and XSS: Practical Examples Using Burp Suite
  30. 10.18653/v1/2024.findings-emnlp.624 · DOI
  31. 2025, Introduction To MCP. Model Context Protocol
  32. Surapaneni, 2025, A2A: A New Era of Agent Interoperability
  33. Chang, 2024, AgentNetworkProtocol (ANP) GitHub Repository
  34. 2024, Agent Communication Protocol (Linux Foundation)
  35. 10.1007/978-3-030-58951-6_24 · DOI · Paper page
  36. 10.18653/v1/2025.acl-long.476 · DOI
  37. 10.52202/079017-1421 · DOI
  38. 10.1109/SP46214.2022.9833644 · DOI · Paper page
  39. 10.14722/ndss.2025.230478 · DOI
  40. 2025, JFrog and Hugging Face Join Forces to Expose Malicious ML Models
  41. 10.1145/501978.501980 · DOI · Paper page
  42. 10.18653/v1/2022.emnlp-main.425 · DOI
  43. 10.18653/v1/2022.findings-emnlp.148 · DOI · Paper page
  44. 10.1145/2619091 · DOI
  45. Singh, 2025, LLM-Based Agents: The Benefits and the Risks
  46. Chhabra, What data benefits my classifier? Enhancing model performance and interpretability through influence-based data selection
  47. 10.52202/075280-0056 · DOI
  48. Petrovic, 2025, Advanced Interpretability Techniques for Tracing LLM Activations
  49. 10.52202/079017-0664 · DOI
  50. de Castro, Privacy-preserving large language model inference via gpu-accelerated fully homomorphic encryption
  51. Lu, 2024, An efficient and extensible zero-knowledge proof framework for neural networks, Cryptol. ePrint Arch., 2024/703
  52. 10.1163/2210-7975_hrd-9919-2016005 · doi:10.1163/2210-7975_hrd-9919-2016005
  53. Cihon, Chilling autonomy: Policy enforcement for human oversight of AI agents
  54. 10.1063/5.0222987 · DOI
  55. 2024, Fine-Tuned DeBERTa-V3-Base for Prompt Injection Detection
  56. Chen, 2024, Aligning LLMs to be robust against prompt injection, arXiv:2410.05451 · arXiv
  57. Helff, Llavaguard: VLM-based safeguard for vision dataset curation and safety assessment, 8322
  58. 10.1609/aies.v7i1.31664 · DOI
  59. 2024, CloudImposer: Executing Code on Millions of Google Servers With a Single Malicious Package
  60. 2025, NCCoE Cyber AI Profile
  61. 2025, Agentic AI—Threats & Mitigations
  62. 10.1007/978-3-032-02130-4_2 · DOI
  63. 10.18653/v1/2025.findings-naacl.123 · DOI
  64. 10.52202/079017-1650 · doi:10.52202/079017-1650 · Paper page
  65. 10.18653/v1/2024.acl-long.50 · DOI · Paper page
  66. 10.18653/v1/2025.findings-acl.1113 · DOI
  67. 10.18653/v1/2025.acl-industry.75 · DOI
  68. 10.18653/v1/2025.acl-long.1575 · DOI
  69. 10.18653/v1/2025.findings-acl.332 · DOI
  70. 10.18653/v1/2025.findings-acl.349 · DOI

Cited by

See all 30

Metadata

Authors
Chhabra, A., Datta, S., Nahin, S. K., & Mohapatra, P.
Affiliations
University of South Florida (US)
Venue
IEEE Access
Published
2026
Volume
14
Pages
49455-49482
ISSN
2169-3536 (ISSN-L)
OpenAlex
W7138839765
Cited by
28
References
288
Citations by year
2026: 28
FWCI
95.69
Type
journal article
Language
en
Open access
Locations
  • IEEE Access (publishedVersion, cc-by)
  • arXiv (Cornell University) (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • arXiv (Cornell University)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Adversarial Robustness in Machine Learning
  • Physical Sciences › Computer Science › Hardware and Architecture › Security and Verification in Computing
  • Physical Sciences › Computer Science › Information Systems › Information and Cyber Security
Keywords
agentic AI, large language models, AI security, security threats, threat taxonomy, security by design, AI governance, AI agent evaluation, autonomous agents, tool use, AI safety

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.