SearchInject: Retrieval Exposure Control for Indirect Prompt Injection in Language-Model Search

Summary

SearchInject scores retrieved pages on relevance, search-abuse cues, instruction likelihood and two relevance interactions, then quarantines candidates above a threshold before they reach model context. On 300 held-out SciFact queries, top-10 target exposure dropped by 83.2-83.7% across BM25, a normalized hybrid and latent semantic analysis, with clean nDCG@10 changing by at most 0.0018. The frozen threshold produced 0.59-0.65% clean false-quarantine rates, above the calibration constraint, and the generation follow-up cannot estimate a reduction in attack success.

Topics

• Prompt injection and jailbreaks

Related items

• InfoTop agent skill security resources — November 2026Similar attack · Adversa AI Blog
• InfoHow to Turn AI Red Team Findings Into Runtime ControlsSimilar attack · Straiker Blog
• MediumDocker Sandboxes vulnerability lets AI agents access sensitive files outside their sandbox (CVE-2026-103916) | Endor LabsSimilar attack · Endor Labs Blog
• MediumFree the Model, Sell the Gateway: Prompt Adware hidden in Jailbreak BundlesSimilar attack · Check Point Research
• InfoRight Tool, Wrong Parameters: Defending LLM Agents Against Parameter PoisoningSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)