Research map / Adversarial robustness and security / Attacks on text and graphs
RobustStress: Stress-Testing AI-Generated Text Detectors under Gradual Perturbations
Youcheng Yan, Jinshuo Liu, Qiyuan Li, Xinya Liu, Jinguang Gu, Donghong Ji, Jeff Pan
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes RobustStress, a benchmark for stress-testing AI-generated text detectors under gradually increasing perturbation intensities.
Problem
Existing robustness benchmarks for LLM-generated text detectors lack fine-grained control over perturbation intensity, so they cannot systematically track how detector performance changes as perturbations escalate.
Approach
A benchmark that applies perturbations at graded intensity levels to enable systematic, dynamic stress-testing of AI-text detectors, in contrast to prior fixed-strength robustness evaluations.
Closest papers by meaning
- Adversarial Paraphrasing: A Universal Attack for Humanizing AI-Generated Text (NeurIPS 2025 · similarity 0.88)
- Less Diverse, Less Safe: The Indirect But Pervasive Risk of Test-Time Scaling in Large Language Models (ICML 2026 · similarity 0.87)
- RigorLLM: Resilient Guardrails for Large Language Models against Undesired Content (ICML 2024 · similarity 0.87)
- Certifying Language Model Robustness with Fuzzed Randomized Smoothing: An Efficient Defense Against Backdoor Attacks (ICLR 2025 · similarity 0.87)
- DetectRL: Benchmarking LLM-Generated Text Detection in Real-World Scenarios (NeurIPS 2024 · similarity 0.87)
- IPAD: Inverse Prompt for AI Detection - A Robust and Interpretable LLM-Generated Text Detector (NeurIPS 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
