Search papers, labs, and topics across Lattice.
The paper introduces Many-Tier Instruction Hierarchy (ManyIH), a new paradigm for resolving instruction conflicts in LLM agents with arbitrarily many privilege levels, addressing the limitations of existing fixed-hierarchy approaches. To evaluate this, they created ManyIH-Bench, a benchmark with up to 12 levels of conflicting instructions across 853 agentic tasks. Experiments on frontier models reveal poor performance (~40% accuracy) on ManyIH-Bench, highlighting the challenge of scalable instruction conflict resolution.
Even state-of-the-art LLMs struggle to follow complex instruction hierarchies, achieving only ~40% accuracy when navigating conflicts across a dozen privilege levels in agentic tasks.
Large language model agents receive instructions from many sources-system messages, user prompts, tool outputs, other agents, and more-each carrying different levels of trust and authority. When these instructions conflict, agents must reliably follow the highest-privilege instruction to remain safe and effective. The dominant paradigm, instruction hierarchy (IH), assumes a fixed, small set of privilege levels (typically fewer than five) defined by rigid role labels (e.g., system>user). This is inadequate for real-world agentic settings, where conflicts can arise across far more sources and contexts. In this work, we propose Many-Tier Instruction Hierarchy (ManyIH), a paradigm for resolving instruction conflicts among instructions with arbitrarily many privilege levels. We introduce ManyIH-Bench, the first benchmark for ManyIH. ManyIH-Bench requires models to navigate up to 12 levels of conflicting instructions with varying privileges, comprising 853 agentic tasks (427 coding and 426 instruction-following). ManyIH-Bench composes constraints developed by LLMs and verified by humans to create realistic and difficult test cases spanning 46 real-world agents. Our experiments show that even the current frontier models perform poorly (~40% accuracy) when instruction conflict scales. This work underscores the urgent need for methods that explicitly target fine-grained, scalable instruction conflict resolution in agentic settings.
