MobileInject Cross-Modal Consistency Injection Against Mobile Agents

# MobileInject Cross-Modal Consistency Injection Against Mobile Agents

https://gcdradar.vercel.app/incidents/mobileinject-cross-modal-consistency-injection-against-mobil · AIR-2026-1396 · 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: MobileInject, Mobile MLLM Agents

## Summary

Researchers presented MobileInject to evaluate how multimodal large language model mobile agents are vulnerable to indirect cross-modal prompt injection. The attack embeds malicious instructions within cross-application smartphone content to steer agent behavior.

## What to do

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

## Latest updates

- 2026-10-19: MobileInject: Evaluating Reasoning-to-Action Security of MLLM-Based Mobile Agents under Cross-Modal Consistency Injection (https://arxiv.org/abs/2610.23307)

## Sources

- 2026-10-19 arXiv: MobileInject: Evaluating Reasoning-to-Action Security of MLLM-Based Mobile Agents under Cross-Modal Consistency Injection https://arxiv.org/abs/2610.23307

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