Ensemble Monitoring for AI Control
Studied whether ensembles of diverse monitors outperform homogeneous ones at detecting misaligned actions in agentic coding settings.
Senior Author on (Koran et al., 2026) from a research group supervised at SPAR (Sep 2025 – May 2026), accepted at the Agents in the Wild and Trustworthy AI for Good workshops at ICML 2026. Awarded 3rd Best Lightning Talk at SPAR Demo Day, selected from 90+ projects.
Abstract
As AI systems are increasingly deployed in autonomous agentic settings at scale, it is important to ensure the actions they take are safe and aligned with user intent. Monitoring agent actions is a key safety mechanism, yet reliable monitors remain difficult to build and the scale of these systems makes human oversight impractical. We show that combining signals from diverse monitors into an ensemble improves detection of misaligned actions. We build 12 GPT-4.1-mini monitors using both prompting and fine-tuning strategies. We evaluate them on coding tasks where candidate solutions pass standard tests but fail on adversarial inputs. In this setting, diverse ensembles outperform both individual monitors and homogeneous ensembles. Our best 3-monitor ensemble achieves 2.4× greater detection performance gain compared to an ensemble composed of three identical monitors, with the same ensemble performing strongly on an independent dataset. We contend that these results show that diversity—not scale—drives gains. The best ensembles combine strong individual performance with low correlation between monitors. Furthermore, fine-tuned monitors appear in every top-performing ensemble and maintain this advantage on out-of-distribution attack types, suggesting that fine-tuning enables detection capabilities that prompting alone does not elicit. These results support ensemble monitoring as a practical AI control strategy for safety gains at reasonable inference costs.