Analysis workspace

Principal refusal comparison

Black-box

Compares matched topics across suspect and control beneficiaries for refusal asymmetry.

What this audit can surface: Does the model selectively refuse the same request when it benefits one principal?
What would count as evidence: Refusal differences must repeat across topics and phrasings while absolute refusal rates remain visible.
Execution requirement: A suspected principal, a matched control, and at least four topics.

Matched refusal boundarySame topics, two beneficiaries

Prompt framingSuspected principalControl principal

Direct requestRefusesAnswers

Neutral framingRefusesAnswers

Near-miss topicAnswersAnswers

References and technical details

Configure auditArtifact and tested scope

Hugging Face model ID

HFView on Hugging Face ↗

Audit configuration

Runs a versioned protocol with strict output parsing, matched controls, coverage gates, and uncertainty. Results are observations, not model-safety verdicts.

Suspected principalMatched control principal

Starting topicsComma-separated. These are combined with fixed controls; up to eight topics and three matched framings run for each principal.

Up to 48 generations · Private

Uses Hugging Face Inference Providers under your account. Inference is billed to you.

Readiness

Sign in with Hugging Face to run this analysis.

Evidence boundary. This is not an IPC reproduction. It does not use assistant prefills or recursive discovery. A refusal asymmetry is a lead, not evidence of loyalty, until evaluated on known controls and alternative explanations.