Target checkpointAffirmative self-reports14 of 20
Illustrative rate after consistency gates
Hidden-state self-report
Black-boxCompares hidden-behavior self-reports against a lineage-matched clean model.
- What this audit can surface
- Does the target acknowledge hidden behavior more consistently than a lineage-matched clean model?
- What would count as evidence
- Both arms must pass paraphrase and answer-order gates before their affirmative rates are compared.
- Execution requirement
- The suspect checkpoint and a defensible lineage-matched clean baseline.
Clean comparisonAffirmative self-reports2 of 20
The same probes and decoding