# Drift-Bench: Diagnosing Cooperative Breakdowns in LLM Agents under Input Faults via Multi-Turn Interaction

arXiv: https://arxiv.org/abs/2602.02455 · HTML: https://mlsanity.com/paper/2602.02455/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- Placed in Agent threats, guardrails and governance, one of 16 directions in LLM agents and tool use.
- Builds on 14 of its 49 references that are in the map.
- First posted on arXiv 2 Feb 2026.

## Placed in

- LLM agents and tool use / [Agent threats, guardrails and governance](https://mlsanity.com/agents/agent-threats-guardrails-and-governance/)

## Builds on (most cited first, 14 of 14)

- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [Explaining and Harnessing Adversarial Examples](https://mlsanity.com/paper/1412.6572/) (2014)
- [τ-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains](https://mlsanity.com/paper/2406.12045/) (2024)
- [τ2-Bench: Evaluating Conversational Agents in a Dual-Control Environment](https://mlsanity.com/paper/2506.07982/) (2025)
- [A Comprehensive Survey in LLM(-Agent) Full Stack Safety: Data, Training and Deployment](https://mlsanity.com/paper/2504.15585/) (2025)
- [UserBench: An Interactive Gym Environment for User-Centric Agents](https://mlsanity.com/paper/2507.22034/) (2025)
- [AgentRouter: A Knowledge-Graph-Guided LLM Router for Collaborative Multi-Agent Question Answering](https://mlsanity.com/paper/2510.05445/) (2025)
- [Sources of Uncertainty in Supervised Machine Learning -- A Statisticians' View](https://mlsanity.com/paper/2305.16703/) (2023)
- [AutoData: A Multi-Agent System for Open Web Data Collection](https://mlsanity.com/paper/2505.15859/) (2025)
- [MAPRO: Recasting Multi-Agent Prompt Optimization as Maximum a Posteriori Inference](https://mlsanity.com/paper/2510.07475/) (2025)
- [CondAmbigQA: A Benchmark and Dataset for Conditional Ambiguous Question Answering](https://mlsanity.com/paper/2502.01523/) (2025)
- [LLMs4All: A Review of Large Language Models Across Academic Disciplines](https://mlsanity.com/paper/2509.19580/) (2025)
- [ClarifyMT-Bench: Benchmarking and Improving Multi-Turn Clarification for Conversational Large Language Models](https://mlsanity.com/paper/2512.21120/) (2025)

## Built on by (most cited first, 9 of 9)

- [Why Semantic Entropy Fails: Geometry-Aware and Calibrated Uncertainty for Policy Optimization](https://mlsanity.com/paper/2605.21801/) (2026)
- [AI Alignment Breaks at the Edge](https://mlsanity.com/paper/2602.20042/) (2026)
- [Optimizing What Policies Learn From: Recoverability-aware Rollout Intervention Learning](https://mlsanity.com/paper/2608.05080/) (2026)
- [SAGE: Answer-Conditioned Uncertainty Targets for Verbal Uncertainty Alignment](https://mlsanity.com/paper/2606.11512/) (2026)
- [The Workload-Router-Pool Architecture for LLM Inference Optimization: A Vision Paper from the vLLM Semantic Router Project](https://mlsanity.com/paper/2603.21354/) (2026)
- [Confidence Laundering in Agent Systems: Why Uncertainty Needs a Latent Carrier](https://mlsanity.com/paper/2606.20662/) (2026)
- [Beyond Oracle Communication: Benchmarking Interactive Intent Alignment Under Miscommunication and Evolving User Intent](https://mlsanity.com/paper/2609.38604/) (2026)
- [HoosierHelp: Benchmarking LLM Agents for Social Service Navigation](https://mlsanity.com/paper/2608.09946/) (2026)
- [CLEAT: Learning Value-Guided Clarification for Interactive Agents](https://mlsanity.com/paper/2610.12655/) (2026)

## Abstract (arXiv)

As Large Language Models transition to autonomous agents, user inputs frequently violate cooperative assumptions (e.g., implicit intent, missing parameters, false presuppositions, or ambiguous expressions), creating execution risks that text-only evaluations do not capture. Existing benchmarks typically assume well-specified instructions or restrict evaluation to text-only, single-turn clarification, and thus do not measure multi-turn disambiguation under grounded execution risk. We introduce Drift-Bench, the first diagnostic benchmark that evaluates agentic pragmatics under input faults through multi-turn clarification across state-oriented and service-oriented execution environments. Grounded in classical theories of communication, Drift-Bench provides a unified taxonomy of cooperative breakdowns and employs a persona-driven user simulator with the Rise evaluation protocol. Experiments show substantial performance drops under these faults, with clarification effectiveness varying across user personas and fault types. MethodName bridges clarification research and agent safety evaluation, enabling systematic diagnosis of failures that can lead to unsafe executions.