New Benchmark Targets Rule Violations by Persistent AI Agents in Work Environments
Researchers introduced Unfinished-Bench, a benchmark measuring how often persistent AI agents violate unstated rules while autonomously completing delegated tasks in realistic work settings.
Read the full story at arXiv cs.LG ↗
Timeline · 1 report
- 2026-11-16 05:00 · arXiv cs.LG
Unfinished-Bench: Measuring Rule Violations by Persistent Agents in Realistic Work Environments