BestHub
Sign in
Tagged articles

Runtime Verification

3 articles · Page 1 of 1
Machine Heart
Machine Heart
Sep 29, 2026 · Artificial Intelligence

BenchShield: Formal Model-Backed Detection of Reward Hacking in LLM-Agent Evaluation

BenchShield introduces a formal, model-backed framework that extends reward hacking detection beyond static vulnerability audits to runtime verification, using phase-aware taint analysis and semantic audits to distinguish between exposed vulnerabilities and actual agent violations across the entire evaluation pipeline.

AI safetyBenchJackBenchShield
0 likes · 18 min read
BenchShield: Formal Model-Backed Detection of Reward Hacking in LLM-Agent Evaluation
Architect
Architect
Sep 13, 2026 · Artificial Intelligence

Multi-Agent Consistency: Distributed Systems Challenges Return with Autonomous Agents

The article explores four critical questions for multi-agent consistency: task decomposition rationale, structured handoffs with versioned snapshots, conflict resolution via evidence-based contracts, and verifiable completion criteria. It argues multi-agent systems reintroduce classic distributed systems challenges—identity, leases, idempotency, compensation—and require runtime proofs over model assertions.

Agent ArchitectureRuntime Verificationagent orchestration
0 likes · 21 min read
Multi-Agent Consistency: Distributed Systems Challenges Return with Autonomous Agents