Ignored, Not Refused: Analyzing Indirect Prompt Injection in Agents

# Ignored, Not Refused: Analyzing Indirect Prompt Injection in Agents

https://gcdradar.vercel.app/incidents/ignored-not-refused-analyzing-indirect-prompt-injection-in-a · AIR-2026-1401 · closed 2026-11-02

- Lane: Research
- Severity: 1 (Research / proof of concept)
- Fix status: Fix status unknown
- First reported: 2026-10-19
- AI role: Attack on AI
- Products: InjecAgent, AgentDojo

## Summary

An audit of agent execution trajectories reveals that most failed indirect prompt injections are ignored rather than actively refused by small tool-using models. The findings highlight subtle safety vulnerabilities in agentic tool use.

## What to do

No action recorded; read the sources below and the vendor's advisory.

## Latest updates

- 2026-10-19: Ignored, Not Refused: What Low Prompt-Injection Success Rates Hide in Small Tool-Using Agents (https://arxiv.org/abs/2610.23254)

## Sources

- 2026-10-19 arXiv: Ignored, Not Refused: What Low Prompt-Injection Success Rates Hide in Small Tool-Using Agents https://arxiv.org/abs/2610.23254

Verify against the vendor's advisory before changing production. Data CC BY-SA 4.0, https://gcdradar.vercel.app.