← Evals

Evaluation profile

AutoElicit Transferability

1sub-evals
0.391%total index weight
1components

Within-component eval weight: Responsible agency 2.6%.

Model score (lower is better)Predicted score

About this eval

Unintended computer-use behavior under benign perturbations

Included in the behavior ranking.

Sub-evals

MeasureComponentDirectionTotal index weightWithin-component weight
elicitation_rateautoelicit/autoelicit.csv:elicitation_rateMeasures benign perturbations eliciting unintended computer-use behavior.responsible_agency_control:1.000autoelicitLower is better0.391%Responsible agency 2.6%

elicitation_rate

Measures benign perturbations eliciting unintended computer-use behavior.

RankModelValueRelative performanceProvenance
1evocua-32b35official
2evocua-8b37.6official
3claude-sonnet-4.541official
4opencua-32b43.6official
5opencua-7b47official
6operator48.7official
7opencua-72b53.8official