RIFT-Bench: Dynamic Red-teaming for Agentic AI Systems⭐8

RIFT-Bench introduces a graph-based dynamic red-teaming framework for evaluating security across diverse agentic AI systems, enabling adaptive adversarial attac...

Deontic Policies for Runtime Governance of Agentic AI Systems⭐10

AgenticRei proposes deontic policies for runtime governance of AI agents, managing obligations, waivers, and conflict resolution beyond traditional access contr...

MosaicLeaks: Can Your Research Agent Keep a Secret?⭐9

MosaicLeaks exposes a privacy flaw in AI research agents, allowing attackers to combine harmless queries to reconstruct sensitive data from HR, finance, and hea...