Evidence · DiggingBeagle record
Judging Before Propagation: Task-Graph Contextual Integrity for Multi-Agent Privacy
The authors report 85.8 to 94.1 percent fewer unauthorized disclosures under task-graph contextual integrity configurations while preserving 93.6 to 96.1 percent of baseline task accuracy.
- Source role
- primary disclosure
Evidence record
The authors report 85.8 to 94.1 percent fewer unauthorized disclosures under task-graph contextual integrity configurations while preserving 93.6 to 96.1 percent of baseline task accuracy.
Claim-level citations (1)
- supportsTask-graph disclosure tracking blocks cumulative multi-agent privacy leakage: The authors report 85.8 to 94.1 percent fewer unauthorized disclosures under task-graph contextual integrity configurations while preserving 93.6 to 96.1 percent of baseline task accuracy.
Paper, MAPLE-Bench and Privacy-Cost Trade-off experiment