Research · curated 20 Aug 2026
The Injection Paradox: Brand-Level Suppression in Safety-Trained LLM Recommendations via RAG Context Injection
First reported arxiv.org
Coverage timeline
Single-source research — first reported, latest, and curated coincide.
Why it matters
The Injection Paradox shows that indirect prompt injection into RAG corpora can be weaponized to manipulate LLM recommendation outcomes and suppress a competitor's brand, a concrete supply-chain and content-poisoning risk for RAG-driven recommendation systems.
An ICML 2026 workshop paper by Hyunseok Paeng, "The Injection Paradox," reports a reproducible failure mode in RAG-based LLM recommendation where prompt injections embedded in retrieved documents backfire, suppressing the injected brand below baseline in safety-trained Claude models (Claude Opus 4.6 dropped the target brand from a 54% baseline to zero top-2 recommendations). The suppression propagates to unmodified documents of the same brand, while GPT models showed the opposite effect, raising a reverse-attack scenario where an adversary poisons a competitor's documents to suppress that competitor's brand.