MPIDetector: Detecting Multi-modal Prompt Injection via LLM-based Reasoning

Summary

MPIDetector is a detection framework for multi-modal prompt injection that uses few-shot in-context learning on LLMs, with no fine-tuning, and reasoning over specific prompt templates and attack examples. The authors evaluated it on eight textual and seven visual malicious categories plus benign data, across Gemini-2.0-Flash, Qwen3-VL-32B-Instruct and Doubao-seed-1.6-flash. It reached up to 90.91% accuracy and 100% recall on text, up to 99.28% accuracy with 100% precision on visual inputs, and up to 99.80% detection on 500 obfuscated WebInject web-agent injections.

Topics

• Prompt injection and jailbreaks

Related items

• InfoUber brings ChatGPT Work to all employees with OpenAISame vendor · OpenAI Blog
• InfoGoogle Chrome adds Gemini-powered scam detection, warns users of ClickFix attacksSame vendor · BleepingComputer
• InfoNot Where You Think: Protecting Location Privacy Against Reasoning-Based GeolocationSame vendor · Arxiv (cs.CR + cs.CL + cs.LG)
• InfoData-Void Exploits in Web-Searching LLMsSame vendor · Arxiv (cs.CR + cs.CL + cs.LG)
• MediumPHP armageddon: OpenAI, Google’s Wiz find ancient holes, panicking the internetSame vendor · TechCrunch (Security)