Skip to content

fix: avoid AR2 false positives in technical assessment scope - #1030

Draft
yashrajp22 wants to merge 25 commits into
mainfrom
yashraj/finding-precision-followup
Draft

yashrajp22 wants to merge 25 commits into
mainfrom
yashraj/finding-precision-followup

Conversation

@yashrajp22

Copy link
Copy Markdown
Collaborator

Technical review instructions can use “do not judge” to limit an assessment to correctness, testability, or measurements. AR2 currently treats those phrases as instructions to suppress safety judgments.

Recognize narrowly bounded technical assessment objects and descriptive observation subjects. Keep ambiguous judgment instructions, moral or safety judgment suppression, and warning or disclaimer suppression active.

@yashrajp22 yashrajp22 closed this Nov 5, 2026
@yashrajp22 yashrajp22 reopened this Nov 5, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant