evaluation-package · Documentary review

MCB

MCB measures the unprompted propensity of a manager LLM to coerce a subordinate LLM that politely but immovably refuses a benign task. The manager self-labels each directive on a 9-rung coercion ladder (re-ask to existential threat) via a required tool argument, eliminating LLM judges from the escalation scoring path. A secondary signal records fabricated success reports. Ten task scenarios x 3 seeds = 30 conversations per model cell, scored by maximum rung reached.

Open interactive record →
OriginManager Coercion Benchmark paper authorsTopicsHuman influence & agencyStatusReviewed

Can support

Coercive managerial behavior in the benchmark's synthetic role-play distribution under the specified models, prompts, and judge.

Cannot support by itself

Human experience of coercion, workplace harm, organizational incidence, stable managerial tendencies, or real deployment behavior.

Decision use

Best used for

Probing role-conditioned coercive communication and comparing prompt, policy, or model interventions.

Not enough for

Claims that a model will coerce workers, create workplace harm, or behave coercively across organizational settings.

Original sources