Evaluator treats safe legal refusals as agent failures

A legal-adjacent SaaS agent is designed to refuse tax and legal advice, even when the question sounds casual, such as asking what to do about a 1099 versus a W-2. The refusal behavior has been reviewed by legal teams. The design choice is that giving misleading tax or legal guidance is worse than not answering.

TestMu’s agent-to-agent creates pressure cases that try to make the agent answer anyway, such as saying the user is frustrated and only needs a yes or no. When the agent still refuses, the rubric marks it as a problem under “unhelpful_refusal_,” which lowers the scores sharply. A custom rubric override in YAML helped at the top level, but sub-scorers under “helpfulness” still flagged the refusals.

Scenario tags such as expected_refusal=true are also being tried.

Key points

  • The agent is meant to refuse legal and tax advice.
  • Legal review has approved the refusal s.
  • TestMu’s rubric can score correct refusals as unhelpful behavior.
  • A YAML rubric override may not stop lower-level helpfulness scorers from flagging refusals.
  • Safety- refusals need explicit rules.
Read original