This report synthesises findings from 10 peer-reviewed papers addressing the following research question: How does the sample efficiency of Mul-GAD in low-label regimes compare to recent contrastive learning approaches for graph anomaly detection. Anomaly detection has been used for decades to identify and extract anomalous components from data. Many techniques have been used to detect anomalies. 9 claims were extracted from source literature; 9 were independently verified against retrieved documents. An automated multi-reviewer quality assessment produced a score of 8.0/10. This report is a machine-generated literature synthesis and does not constitute original research.Research goal: How does the sample efficiency of Mul-GAD in low-label regimes compare to recent contrastive learning approaches for graph anomaly detection?Autonomous literature synthesis. Automated review score: 8.0/10. Full text and citation available at Assignee Research.