paper
Efficient and Sound Probabilistic Verification for AI Agents
paperactiveprovisional
efficient-and-sound-probabilistic-verification-for-ai-agents-e037a3d9·1 events·first seen 47h agoAliases: Efficient and Sound Probabilistic Verification for AI Agents
Co-occurring entities
More like this (12)
Towards a Science of AI Agent ReliabilityConcrete Problems in AI SafetyVisual Verification Enables Inference-time Steering and Autonomous Policy ImprovementTrustworthy AIProvenanceGuard: Source-Aware Factuality Verification for MCP-Based LLM AgentsBayesian Inference and Decision Audits for Public Archives of Frontier AI EvaluationsLearning Red Agent Policy from Observations for Neurosymbolic Autonomous Cyber Agentsspeculative execution (AI agents)third-party AI evaluationsAI-assisted theorem provingMulti-Turn Evaluation of Deep Research Agents Under Process-Level FeedbackAPPO: Agentic Procedural Policy Optimization
Recent events (1)
Distributionally robust optimization framework for probabilistic runtime verification of AI agents
A new arXiv preprint introduces a sound and efficient framework for verifying probabilistic security policies for AI agents operating in complex digital environments, addressing limitations of prior Datalog-based approaches that assumed deterministic policies or predicate independence. The method uses distributionally robust optimization to compute sound upper bounds on policy violation probability without requiring independence assumptions between predicates. Evaluated on benchmarks for terminal and tool-calling agents, the approach outperforms prior art on the security-utility trade-off.