When Actions Go Off-Task: Detecting and Correcting Misaligned Actions in Computer-Use Agents
Yuting Ning, Jaylen Jones, Zhehao Zhang, Chentao Ye, Weitong Ruan, Junyi Li, Rahul Gupta, Huan Sun · 2026
Read When Actions Go Off-Task: Detecting and Correcting Misaligned Actions in Computer-Use Agents by Yuting Ning, Jaylen Jones, Zhehao Zhang, Chentao Ye, Weitong Ruan, Junyi Li, Rahul Gupta, Huan Sun (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-08995.txt.
Description
Computer-use agents that click and type on your behalf sometimes drift off-task, through outside manipulation or their own faulty reasoning. Yuting Ning and seven co-authors define and study misaligned-action detection for the first time, covering both external attacks like indirect prompt injection and internal slips. They build MisActBench, a benchmark of realistic trajectories with human action-level alignment labels.
Their guardrail, DeAction, checks each action before it runs and corrects it through structured feedback. It beats all baselines by over fifteen F-one points on the benchmark. In live tests it cuts attack success by over ninety percent while preserving or improving task success in benign settings, at moderate latency cost.
Catalog record
| CREATOR | Yuting Ning, Jaylen Jones, Zhehao Zhang, Chentao Ye, Weitong Ruan, Junyi Li, Rahul Gupta, Huan Sun |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 75 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-08995