Preprint Open access
When a Kindergartener Solves Calculus: Measuring Capability Leakage in Role-Prompted Reasoning Models
We investigate the problem of role-capability leakage (RCL), in which a role-prompted reasoning model generates convincing in-role text while continuing to exhibit capabilities on benchmarks that exceed those implied by the assigned role. For example, when a model is prompted to assume the role of a kindergarten studen …