cd/entity/STEER-Bench· home› entities› STEER-Bench
grep -l @steer-bench /news/*.json | wc -l → 1

STEER-Bench

mentions 1 type Organization feed RSS

// recent coverage 1 mentions

09:07
2026-10-05
arxiv.org
ai-safety

Securing Computer-Use Agents Against Branch Steering Attacks

A 2 October 2026 arXiv paper introduces STEER-Bench, a 101-task benchmark across 9 domains, and reports that branch steering attacks succeed against 94.4% of standard Computer Use Agents and 89.5% of …

// co-occurs with top 6 entities
COBRA 1 Computer Use Agents 1 Dual-LLM 1 Planner LLM 1 Quarantined LLM 1 arXiv 1
// topics top 5 topics
ai safety 1 ai agents 1 ai research 1 large language models 1 artificial intelligence 1