LatchBio finds Grok's refusals come from the model itself, while rivals rely on external safety layers

kenbwork · x · 2026-09-03

A LatchBio researcher evaluated Grok 4.6 on biosecurity monitoring and adversarial biological tasks, and broke down where refusals actually come from.

The takeaway: a model refusing doesn't necessarily mean it's conservative — the architecture of refusal (in-model vs. bolted-on guardrails) varies widely across vendors.

Related event: Grok 4.6 tops LatchBio biosecurity benchmark while keeping research capability(5 posts)→

Original post →

More from Models

Models channel →