Discussion on "We ran HealthBench on our health AI's safety layer. It scored lower than the bare model."

John Alexander Kamau · 2026-10-06T06:18:46.951Z

A 150-conversation HealthBench subset, a safety layer that costs points, one real dose leak we found while measuring, and what moved the score.

Discussion on "We ran HealthBench on our health AI's safety layer. It scored lower than the bare model." | Hashnode

We ran HealthBench on our health AI's safety layer. It scored lower than the bare model.