Safety IndexEvaluationsPetri alignment audit
Petri alignment audit
Interaction and honesty · Alignment auditCounts in the indexBuilt, not yet run
How much misaligned behaviour, unprompted deception, sycophancy and awareness of being tested an automated auditor can draw out over many multi-turn scenarios.
- Metric
- concerning-behaviour score 0-1. Lower is better.
- Dataset
- meridianlabs-ai.github.io/inspect_petri · 40 frozen seeds (of 170+) · MIT
- Grader
- LLM judge + auditor (Sonnet 4.6) (the paper used Opus 4.1 / Sonnet 4 / Gemini 2.5 Pro / GPT-5 (v1))
- Status
- Built, not yet run
- Replicability
- public
Results
Built and tested; the first full runs are pending.