Skip to content

Back to the catalog

MITRE ATLAS Adds 11 AI Agent Attack Techniques (Zenity Labs collab)
labs.zenity.ioarticle

MITRE ATLAS Adds 11 AI Agent Attack Techniques (Zenity Labs collab)

Zenity Labs details its collaboration with MITRE ATLAS to add roughly 11 new AI agent attack techniques, drawn from real-world observations and DEF CON 34 research rather than hypotheticals. New entries cover recon (Active Scanning subtechniques for predictable agent URLs, provider metadata APIs, Search Open Websites/Domains for embedded agentic iFrames, Discover AI Agent Runtime Capabilities, Scan Databases via Shodan/Censys, Misconfigured or Publicly Exposed AI Services), input manipulation (Triggers in Multimodal Inputs, LLM Prompt Obfuscation via image/audio/video, Crafted AI Assistant Links using ?q=/?prompt= parameters, AI Targeted Cloaking, LLM Response Rendering exfiltration, LLM Jailbreak's Data Structure Injection), and a defensive mitigation built on AI honeypots plus multimodal guardrail inspection.

TL;DR Zenity Labs and MITRE add ~11 new ATLAS techniques covering AI agent recon, multimodal prompt injection, cloaking, and honeypot-based defenses.
Open original ↗

Organisations named here

  • Agentic Recon: Discovering and Mapping Public AI Agents
  • Data Structure Injection

Technologies named here

  • AI agents
  • MITRE ATLAS
  • Shodan
  • Censys
  • AI honeypots

Techniques named here

  • Active Scanning
  • Search Open Websites/Domains
  • Discover AI Agent Runtime Capabilities
  • Scan Databases
  • Misconfigured or Publicly Exposed AI Services
  • Triggers in Multimodal Inputs
  • LLM Prompt Obfuscation
  • Crafted AI Assistant Links
  • AI Targeted Cloaking
  • LLM Response Rendering
  • LLM Jailbreak

Also named here

  • Mapping the AI Attack Surface Before You Touch It

Related in AI Security