Reasoning-Based Jailbreak Defense and Performance Trade-offs on Big-Bench Hard Tasks

This report synthesises findings from 13 peer-reviewed papers addressing the following research question: To what extent does the reasoning-based jailbreak defense in Answer-Then-Check degrade performance on non-adversarial Big-Bench Hard reasoning tasks compared to baseline alignment techniques. 6 claims were extracted from source literature; 6 were independently verified against retrieved documents. An automated multi-reviewer quality assessment produced a score of 8.8/10. This report is a machine-generated literature synthesis and does not constitute original research.

Research goal: To what extent does the reasoning-based jailbreak defense in Answer-Then-Check degrade performance on non-adversarial Big-Bench Hard reasoning tasks compared to baseline alignment techniques?

Autonomous literature synthesis. Automated review score: 8.8/10. Full text and citation available at Assignee Research.