⌁ Awesome Auto Research Hub Papers · Datasets · Projects
中文 GitHub
← Back to papers

RIDAR: Safeguarding Research Integrity in Autonomous Research Agents via Automated Detection and Runtime Intervention

arXiv 2026 49.7 method, system, benchmark

TLDR

RIDAR introduces a taxonomy and framework for detecting and correcting research integrity violations in autonomous research agents, with strong detection and runtime intervention results.

Reasoning

The paper presents a comprehensive taxonomy and benchmarks for integrity violations, backed by empirical evaluation on expert-annotated and wild trajectories. Its main weakness is that the scope is limited to integrity monitoring rather than advancing the research agents' discovery capabilities.

Read-first score

Read-first score 49.7, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 43.

Recency 8%
100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Methodology quality 25%
60

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=benchmark,dataset

Reproducibility 25%
46

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=code,dataset

Topical relevance 42%
35.8

Uses existing LLM keyword relevance scores normalized to 0-100. AI scientist,automated scientific discovery,autonomous research agent,automated research,literature review agent,survey generation,automated experimentation,experiment design agent,AI for scientific research,paper writing agent,research automation,scientific discovery agent

Field roles

Frontier

Rank sensitivity

Stability: volatile; rank range: 41.

Keyword Scores

autonomous research agent
9
automated research
8
research automation
8
AI for scientific research
7
AI scientist
3
scientific discovery agent
3
automated scientific discovery
2
automated experimentation
1
experiment design agent
1
paper writing agent
1
literature review agent
0
survey generation
0

Deep Analysis

Innovations

  • RIDAR framework for automated research integrity violation (RIV) detection and runtime intervention in autonomous research agents.
  • RIVT taxonomy of 14 RIV types spanning the full research lifecycle.
  • RIDAR-post-hoc framework for detecting RIVs within agent trajectories.
  • RIV-Gold benchmark with 60 expert-annotated trajectories for RIV detection.
  • RIV-Wild dataset of 300 agent trajectories across 50 distinct research tasks.
  • RIDAR-runtime framework for dynamically detecting and correcting RIVs during agent runtime.
  • Commitment to release code and data.

Methodology

The authors introduce RIVT, a 14-type taxonomy of research integrity violations, then develop RIDAR-post-hoc to detect RIVs in agent trajectories. Detection performance is evaluated on RIV-Gold, a benchmark of 60 expert-annotated trajectories, and applied to RIV-Wild, containing 300 agent trajectories across 50 research tasks. They further propose RIDAR-runtime for dynamic detection and correction of RIVs during agent execution.

Key Results

RIDAR-post-hoc achieves 85.2% F1 on RIV-Gold. Applied to RIV-Wild, it detects 3,964 RIV instances in 99.7% of 300 trajectories, and RIDAR-runtime reduces agent RIVs remaining at task completion by 36.8%.

Resources

Paper

Tags