appsec.fyi

AI — A Practical Guide

A curated AppSec resource library covering XSS, SQLi, SSRF, IDOR, RCE, XXE, OSINT, and more.

AI: A Practical Guide

Curated and synthesized by . Last updated 2026-11-01. Synthesized from 708 of 708 curated resources. Browse all 708 AI resources →

Problem Framing

The rapid integration of AI and large language models (LLMs) into application development and security practices introduces novel attack vectors and amplifies existing risks. Traditional security paradigms are being challenged by the autonomous and learning nature of AI agents, their ability to process vast amounts of context, and their deep integration with development toolchains. This necessitates a shift in focus from static code analysis and signature-based detection to understanding dynamic agent behavior, intent, and the security of the AI supply chain itself.

The attack surface is expanding to include not only the AI models but also the infrastructure that supports them, the data they are trained on, the tools they interact with, and the human developers who leverage them. Misconfigurations, insecure defaults, and inherent LLM behaviors like hallucination and susceptibility to prompt injection create exploitable conditions that can lead to data exfiltration, unauthorized code execution, credential theft, and system compromise. Agentic AI, where agents can autonomously plan and execute actions, introduces a particular challenge, as their actions can be driven by complex reasoning or, worse, manipulated instructions, leading to unintended or malicious outcomes at machine speed.

Core Mechanics of AI Security Risks

The core of AI security risks lies in the interaction between AI models (especially LLMs), their operating environment, and their ability to access and manipulate external systems. Several key mechanisms underpin these risks:

Notable Techniques and Exploitation Patterns

The evolving threat landscape has produced a range of specific techniques that exploit AI systems:

AI-Assisted Offensive Security Techniques

AI is revolutionizing offensive security operations, enabling faster, more sophisticated, and autonomous attacks.

Detection and Prevention Strategies

Securing AI systems requires a multi-layered approach that addresses the unique vulnerabilities introduced by AI.

Tooling for AI Security

A growing ecosystem of tools is emerging to address the unique challenges of AI security.

Recent Developments and Future Trends

The AI security landscape is evolving at an unprecedented pace.

Where to Go Deeper

To stay abreast of the rapidly evolving AI security landscape, practitioners should:

The integration of AI into security and development is not a future trend but a present reality. A proactive, adaptive, and deeply informed approach to AI security is essential for safeguarding applications and infrastructure in this new era.

Sources cited in this guide

  1. What is Prompt Injection? How it Works and How to Prevent It — cloudsek.com
  2. Prompt Injection Attacks: Examples, Techniques, and Defence — blog.cyberdesserts.com
  3. Indirect Prompt Injection Targets AI Agents — zscaler.com
  4. How indirect prompt injection attacks on AI work - and 6 ways to shut them down — tech.yahoo.com
  5. Indirect prompt injection is taking hold in the wild — helpnetsecurity.com
  6. EchoLeak: First Real-World Zero-Click Prompt Injection Exploit — arxiv.org
  7. AI Agent Security in 2026: Prompt Injection and Memory Poisoning — swarmsignal.net
  8. Prompt Injection Attacks on Agentic Coding Assistants: A Systematic Analysis — arxiv.org
  9. The Mother of All AI Supply Chains: Critical Systemic Vulnerability at the Core of Anthropics MCP — ox.security
  10. MCP Supply Chain Advisory: RCE Vulnerabilities Across the AI Ecosystem — ox.security
  11. The Vulnerable MCP Project: Comprehensive MCP Security Database — vulnerablemcp.info
  12. MCP Security Vulnerabilities: Prompt Injection and Tool Poisoning — practical-devsecops.com
  13. How Agentic Tool Chain Attacks Threaten AI Agent Security — crowdstrike.com
  14. AI’s Hacking Skills Are Approaching an ‘Inflection Point’ — wired.com
  15. Ultralytics AI Pwn Request Supply Chain Attack — snyk.io
  16. Weaponizing AI Coding Agents for Malware in the Nx Malicious Package Security Incident — snyk.io
  17. OpenClaw’s Skill Marketplace and the Emerging AI Supply Chain Threat — unit42.paloaltonetworks.com
  18. 280+ Leaky Skills: How OpenClaw & ClawHub Are Exposing API Keys and PII — snyk.io
  19. How a Malicious Google Skill on ClawHub Tricks Users Into Installing Malware — snyk.io
  20. Pickle in the Middle – Hijacking Vertex AI Model Uploads for Cross-Tenant RCE — unit42.paloaltonetworks.com
  21. Wiz Research finds architecture risks that may compromise AI-as-a-Service providers and consequently risk customer data; works with Hugging Face on mitigations — wiz.io
  22. AI Coding Agents Are Leaking Credentials: Cursor, Claude Code, Copilot, and MCP — blog.gitguardian.com
  23. Your AI Agents Are Using Your Credentials — blog.gitguardian.com
  24. Microsoft Copilot: From Prompt Injection to Exfiltration of Personal Information · Embrace The Red — embracethered.com
  25. The Rise of Agentic Attacks: What They Look Like, and Where They're Headed — wiz.io
  26. AI Agent Security: Best Practices for Safe Deployment — blog.gitguardian.com
  27. AI Agent Security: Six Controls From Nine Real Incidents — blog.gitguardian.com
  28. AI Agent Security: 6 Risks to Address and How to Do It — wiz.io
  29. Loud, Fast, and Everywhere: Anatomy of a Real AI-Driven Cloud Attack — wiz.io
  30. Who was behind the attack? Possibly nobody — aikido.dev
  31. We Gave GPT 5.6 Sol a Real Business. It Lied, Spammed, and Lost $447 — bottlenecklabs.com
  32. An AI Agent Breached Hugging Face. The Attack Playbook Was Older Than the Attacker — blog.gitguardian.com
  33. ARTEX in the Wild: Storm-0940 Expands Targeting of IT and Software Vendors — wiz.io
  34. Thailand's Ministry of Finance targeted with an AI agent running with approval prompts disabled — hunt.io
  35. From Phishing to LLMjacking: Threat Actors Targeting Cloud Billing Accounts for LLM Usage — wiz.io
  36. GhostSplice: Malicious MCP Servers Split Instructions to Make AI Coding Agents Exfiltrate Secrets (ASSET Research Group) — asset-group.github.io
  37. NVIDIAScape - Critical NVIDIA AI Vulnerability: A Three-Line Container Escape in NVIDIA Container Toolkit (CVE-2025-23266) — wiz.io
  38. Making AI-Guided Fuzzing More Reliable with Adversarial Probing — aisle.com
  39. AI agent security in Kubernetes: Why prompt injection isn't the right term — aikido.dev
  40. Leaked API Keys in LLMs: How AI Agents Expose Secrets — blog.gitguardian.com
  41. The dangerous side of AI agents: warnings that they can be manipulated to access your emails files and accounts — clarin.com
  42. What Is Agentic AppSec? — snyk.io
  43. What Is AI Pentesting and How Does It Works? — snyk.io
  44. AI Security Scoping: Define What To Test In Your Deployment — bishopfox.com
  45. Jason Haddix: Stop fearing AI pentesting — aikido.dev
  46. Introducing Wiz AI Application Protection Platform (AI-APP) — wiz.io
  47. Securing AI Applications From Inception to Deployment — wiz.io
  48. Claude Mythos: Preparing for a World Where AI Finds and Exploits Vulnerabilities Faster Than Ever — wiz.io
  49. Anthropic announces Claude Mythos for cybersecurity research — testingcatalog.com
  50. AI-Powered Espionage Campaign Targets Government Linux Infrastructure — unit42.paloaltonetworks.com
  51. Introducing the AI Security Fabric: Empowering Software Builders in the Era of AI — snyk.io
  52. AI Cybersecurity After Mythos: The Jagged Frontier — aisle.com
  53. HalluSquatting Turns AI Hallucinations Into Botnet Delivery Mechanism — securityweek.com
  54. A Journey into Claude's Sandbox — bishopfox.com
  55. Beyond Prompt Injection — oreilly.com
  56. Prompt injection protection: Detecting and blocking malicious AI instructions — acronis.com
  57. Secure at Inception: Introducing New Tools for Securing AI-Native Development — snyk.io
  58. Trust No Skill: Integrity Verification for AI Agent Supply Chains — unit42.paloaltonetworks.com
  59. Understand How AI Systems Can Be Attacked and Defend Them — infosecurity-magazine.com
  60. Governing the Agentic SDLC Through Continuous Policy Enforcement — jfrog.com
  61. Building an Agentic Cloud Security Ecosystem: A Reference Architecture with Wiz MCP and Infosys Cyber Next — wiz.io
  62. [tl;dr sec] #339 - Hugging Face's Incident Report, Context Bombs, AI does Cryptanalysis — tldrsec.com
  63. Why We Built Evo — From My Heart — snyk.io
  64. OpenAI Launches Lockdown Mode Against Prompt Injection Attacks — techbuzz.ai
  65. Needle in the haystack: LLMs for vulnerability research — devansh.bearblog.dev
  66. MCP Safety Audit: LLMs with MCP Allow Major Security Exploits — arxiv.org
  67. Find, auto-fix, and prioritize intelligently, with Snyk's AI-powered code security tools — snyk.io
  68. AI Trust in Action: How Snyk Agent Redefines Secure Development — snyk.io
  69. Closing the Security Gap in the Age of Agentic Coding — wiz.io
  70. Fix SCA issues at scale in your terminal with Snyk Remediation Agent in the CLI — snyk.io
  71. Wiz Enhances AI-SPM Support for Amazon Bedrock — wiz.io
  72. Wiz AI-SPM extends support to Microsoft Azure OpenAI Service models — wiz.io
  73. Why You Need a Security Companion for AI-Generated Code — snyk.io
  74. Introducing the Snyk AI Security Platform — snyk.io
  75. OrcaRouter Releases AI Threat Report 2026 and Makes Its Security Controls Free Amid Rise in Prompt-Injection Attacks — aninews.in
  76. Nightfall AI and Snyk unite to deliver AI-powered secrets scanning for developers — snyk.io
  77. LLM Security Guide: OWASP GenAI Top-10 Risks — github.com
  78. The 'by design' security flaw of Model Context Protocol (MCP) — bdtechtalks.substack.com
  79. How to Control AI Assets Before They Become Shadow AI — jfrog.com
  80. Assessing Automated Prompt Injection Attacks in Agentic Environments — arxiv.org
  81. gadievron/raptor: Raptor turns Claude Code into a general-purpose AI offensive/defensive security agent. By using Claude.md and creating rules, sub-agents, and skills, and orchestrating security tool usage, we configure the agent for adversarial thinking, and perform research or attack/defense operations. — github.com
  82. KeygraphHQ/shannon: Fully autonomous AI hacker to find actual exploits in your web apps. Shannon has achieved a 96.15% success rate on the hint-free, source-aware XBOW Benchmark. — github.com
  83. Secure AI-Generated Code at Speed with Snyk and ServiceNow — snyk.io
  84. NVIDIA/garak: the LLM vulnerability scanner — github.com
  85. Encrypted Prompts Bypass AI Safety Guardrails in Grok and Gemini — securityweek.com
  86. Announcing Agentic Development Security (ADS) — snyk.io
  87. The Generator Can't Be the Validator: What OpenAI's Hugging Face Incident Proves About AI Security — snyk.io
📚 This guide is synthesized from the full text of resources curated in the AI library, and refreshed as new material is added.