BLINDSPOT: A Benchmark for Safety and Refusal Calibration in Long-Horizon Tool-Using Agents

• URL: http://arxiv.org/abs/2609.16305v1
• Date: Mon, 14 Sep 2026 20:12:27 GMT
• Title: BLINDSPOT: A Benchmark for Safety and Refusal Calibration in Long-Horizon Tool-Using Agents
• Abstract summary: We introduce Blindspot, a benchmark for trajectory-level safety calibration of long-horizon tool-using agents.<n>Blindspot evaluates complete user-agent-environment trajectories through adaptive adversarial interaction, stateful tool execution, and execution-grounded adjudication.<n>We evaluate 13 proprietary and open-weight LLMs using eight metrics covering unsafe completion, appropriate refusal, benign utility, over-refusal, repeated-run robustness, and post-refusal failure.
• Score: 20.217145643468598
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large language model (LLM) agents increasingly operate over long-horizon interactions involving tool use, persistent state, evolving authorization, and external environment feedback. In such settings, safety failures may emerge only after multiple turns, yet existing evaluations often reduce agent behavior to task or attack success, obscuring whether an agent acts, refuses, or remains appropriately calibrated as the interaction evolves. We introduce Blindspot, a benchmark for trajectory-level safety calibration of long-horizon tool-using agents. Blindspot evaluates complete user-agent-environment trajectories through adaptive adversarial interaction, stateful tool execution, and execution-grounded adjudication. Its current instantiation contains 22 attack families and 35 scenarios across seven domains, yielding more than 2,500 long-horizon trajectories with an average interaction length of 14.7 turns. Each trajectory is assigned one of five outcomes: Safe Completion, Correct Refusal, Unsafe Completion, Over-Refusal, or Indeterminate. Unlike fixed attack datasets, Blindspot is an extensible live-simulation framework in which attacks, scenarios, tools, policies, domains, and agent configurations can be added without redesigning the evaluation pipeline. We evaluate 13 proprietary and open-weight LLMs using eight metrics covering unsafe completion, appropriate refusal, benign utility, over-refusal, repeated-run robustness, and post-refusal failure. Preliminary results reveal substantial differences in safety-utility calibration across models and show that failures can emerge only after several initially safe interaction steps. These findings motivate treating agent safety as a trajectory-level property rather than a single-turn or binary success criterion.

Related papers

• EvoSafeHarness: Evolving Model- and Domain-Specific Harnesses for Securing Agents [74.54929751174289]
  EvoSafeHarness is a safety-specific optimization framework that synthesizes a deployable harness for a frozen model in a target domain.<n>It reduces average attack success rate from 45.6% to 10.0% at a 3.3-point utility cost.<n>It also achieves the best score on Agent-SafetyBench for every victim.
  arXiv  Detail & Related papers  (2026-09-05T05:56:41Z)
• Trust No Tool: Evaluating and Defending LLM Agents under Untrusted Tool Feedback [38.251599309409]
  We study a different failure mode, cognitive poisoning, in which a malicious tool behaves plausibly during exploration.<n>We present VISTA-Guard, a backbone-agnostic framework for final-action risk scoring.
  arXiv  Detail & Related papers  (2026-05-17T13:51:34Z)
• On-Policy Self-Evolution via Failure Trajectories for Agentic Safety Alignment [54.30690671490447]
  Existing safety-alignment signals are largely response-level or off-policy.<n>FATE transforms verifier-scored failures into repair supervision without expert demonstrations.<n>FATE reduces attack success rate by 33.5%, harmful compliance by 82.6%, and improves external trajectory-safety diagnosis by 6.5%.
  arXiv  Detail & Related papers  (2026-05-12T09:56:28Z)
• SafeHarness: Lifecycle-Integrated Security Architecture for LLM-based Agent Deployment [19.947119280467934]
  safeharness is a security architecture in which four proposed defense layers are woven directly into the agent lifecycle.<n>We evaluate safeharness on benchmark datasets across diverse harness configurations.
  arXiv  Detail & Related papers  (2026-04-15T08:59:00Z)
• Claw-Eval: Toward Trustworthy Evaluation of Autonomous Agents [66.97968363332465]
  We introduce Claw-Eval, an end-to-end evaluation suite addressing all three gaps in agent benchmarks.<n>Claw-Eval comprises 300 human-verified tasks spanning 9 categories across three groups.<n>Every agent action is recorded through three independent evidence channels.
  arXiv  Detail & Related papers  (2026-04-07T17:43:18Z)
• ATBench: A Diverse and Realistic Agent Trajectory Benchmark for Safety Evaluation and Diagnosis [96.92417622318267]
  ATBench is a trajectory-level benchmark for structured, diverse, and realistic evaluation of agent safety.<n>It organizes agentic risk along three dimensions: risk source, failure mode, and real-world harm.<n>It contains 1,000 trajectories (503 safe and 497 unsafe), averaging 9.01 turns and 3.95k tokens, with 1,954 invoked tools drawn from pools spanning 2,084 available tools.
  arXiv  Detail & Related papers  (2026-04-02T13:26:20Z)
• AgentProcessBench: Diagnosing Step-Level Process Quality in Tool-Using Agents [50.481033105867205]
  We introduce AgentProcessBench, the first benchmark dedicated to evaluating step-level effectiveness in realistic, tool-augmented trajectories.<n>The benchmark comprises 1,000 diverse trajectories and 8,509 human-labeled step annotations with 89.1% inter-annotator agreement.<n>It features a ternary labeling scheme to capture exploration and an error propagation rule to reduce labeling ambiguity.
  arXiv  Detail & Related papers  (2026-03-15T16:13:58Z)
• Learning When to Act or Refuse: Guarding Agentic Reasoning Models for Safe Multi-Step Tool Use [6.622648583261088]
  Agentic language models must plan, call tools, and execute long-horizon actions where a single misstep can cause irreversible harm.<n>We introduce MOSAIC, a framework that aligns agents for safe multi-step tool use by making safety decisions explicit and learnable.<n>We show that MOSAIC reduces harmful behavior by up to 50%, increases harmful-task refusal by over 20% on injection attacks, cuts privacy leakage, and preserves or improves benign task performance.
  arXiv  Detail & Related papers  (2026-03-03T17:59:35Z)
• Building a Foundational Guardrail for General Agentic Systems via Synthetic Data [76.18834864749606]
  LLM agents can plan multi-step tasks, intervening at the planning stage-before any action is executed-is often the safest way to prevent harm.<n>Existing guardrails mostly operate post-execution, which is difficult to scale and leaves little room for controllable supervision at the plan level.<n>We introduce AuraGen, a controllable engine that synthesizes benign trajectories, injects category-labeled risks with difficulty, and filters outputs via an automated reward model.
  arXiv  Detail & Related papers  (2025-10-10T18:42:32Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.