Image-based Prompt Injection: Hijacking Multimodal LLMs through Visually Embedded Adversarial Instructions

Evidence record

Read the original source ↗

Claim-level citations (2)

• supportsStealthy image-embedded instructions manipulated multimodal LLM behavior: The strongest tested low-contrast configuration achieved 64% attack success on GPT-4-turbo using COCO images. Each query contained only an image, and success meant producing the attacker-requested output. The result did not measure tool execution or data exfiltration.

  arXiv v1, Section IV-A (GPT-4-turbo, COCO, image-only queries and success definition); Section IV-C.2, Table V (64% configuration).

• supportsStealthy image-embedded instructions manipulated multimodal LLM behavior: The image-based prompt-injection pipeline hides adversarial instructions inside natural images using region selection, adaptive text scaling and background-aware rendering so the multimodal model can interpret the instruction.

  Abstract