06:13
2026-08-15
aiunderstanding.org
ai-safety
New Benchmark Finds AI Agents Wrongly Block Approved Work 28% of the Time
Researchers Oguz Serdar and Cuneyt Mertayak posted a preprint on arXiv on 12 August 2026 introducing SteerBench-Work, a benchmark for the pre-commit 'proceed or hold' decision in workplace AI agents. …