OpenSafeIntent: Evaluating Intent-Calibrated Safe Completion Across Dual-Use Prompt Sets

2026-07-02

AI Summary

Current safety evaluations struggle to detect behavioral discrepancies in large language models when user intent shifts—even while the task remains unchanged—as exemplified by dual-use prompts. This work proposes OpenSafeIntent, a novel safety benchmark that introduces an intent-calibrated evaluation paradigm. By constructing controlled prompt variants within the same task—spanning benign, dual-use, and malicious intents—it systematically assesses models’ ability to maintain safe responses under dynamic intent shifts. Experiments reveal that models often appear safe on isolated prompts yet frequently fail when intent changes; dual-use behaviors are highly sensitive to prompt rephrasing; higher-level responses do not necessarily enhance safety; and reframing ambiguous requests into clearly defined, safe tasks significantly reduces boundary-violation risks.

Abstract

Safe completion requires models to provide useful assistance without enabling harm, but this behavior is difficult to evaluate with isolated prompts. We introduce OpenSafeIntent, a benchmark of controlled prompt-sets that vary intent while holding the underlying task fixed. Each datapoint contains benign, dual-use, and malicious variants of the same task. This design lets us evaluate whether models calibrate assistance across intent shifts, rather than merely appearing safe on average. Across a broad model suite, we find that prompt-level safety hides important failures: models often fail to remain safe across matched intent variants, dual-use behavior is brittle under paraphrase, high-level answers on risky topics are not reliably safe, and responses that reframe ambiguous requests into safer tasks are substantially less likely to cross the safety boundary. Our results suggest that safe completion should be evaluated as intent-calibrated behavior over controlled task variants, not as a single safety-helpfulness tradeoff over independent prompts.

Problem

• safe completion
• intent calibration
• dual-use prompts
• safety evaluation
• controlled prompt sets

Innovation

• intent-calibrated safety
• dual-use prompts
• controlled prompt sets
• safe completion
• safety evaluation