policy
The Verdict Barely Moved When the Rule Changed
Across 20 compliance domains, deleting, swapping or negating the governing rule often left model decisions almost unchanged.
Summary
Across 20 compliance domains, deleting, swapping or negating the governing rule often left model decisions almost unchanged.
Researchers test whether an LLM compliance verdict actually depends on the rule supplied with the case. Across five models and 20 regulatory or platform-policy domains, they delete, swap or negate the rule while holding the facts fixed, then measure changes in the verdict and internal compliance representation. Both measures moved little in many cases. A guard model adapted to custom rules was the least sensitive and least accurate at 51%, while general-purpose models reached 90% to 92%. On harder cases where deleting a rule changed a previously correct prediction, models tracked it more closely. The study shows that accuracy alone cannot establish rule grounding; it does not establish legal reliability for any evaluated system.
Why it matters
Across 20 compliance domains, deleting, swapping or negating the governing rule often left model decisions almost unchanged.
Limits and context
- The study shows that accuracy alone cannot establish rule grounding; it does not establish legal reliability for any evaluated system.
Key claims
Across 20 compliance domains, deleting, swapping or negating the governing rule often left model decisions almost unchanged.
Qualification: The study shows that accuracy alone cannot establish rule grounding; it does not establish legal reliability for any evaluated system.
Evidence: source-2026-10-11-016
Sources
- arXiv preprint 2610.12313, original v1arXiv / submitting research authors · primary research
Corrections
No corrections have been recorded for this story.