Phying News
P非影 中文
Curated security research, vulnerabilities, advisories and tools for practitioners.

Task Injection – Exploiting agency of autonomous AI agents

bughunters.google.com | blog | #ai-security | #prompt-injection | #ai-agents | #google | #task-injection | #autonomous-agents

Summary

This post explains what Task Injection attacks are, how they differ from Prompt Injection, and why autonomous AI agents designed for broad actions are especially exposed.
Published
Collected

original ↗

Related coverage

  1. What’s changed in AI security over the past two years (bugcrowd.com)
  2. Exploiting Tool and Function Calling in LLM Agents (blog.sentry.security)
  3. Google ARTEMIS: Natural-Language Driven Android Automation & Mobile AI Agent Framework (github.com)
  4. StrikeAgent AtkBrain-Flash: Autonomous AI Penetration Testing & Red Teaming Platform (github.com)
  5. Flounder: Autonomous AI White-Hat Security Auditor with Execution-Grounded Verification (github.com)
  6. Google Bug Hunters (bughunters.google.com)

back