AI Engineer · October 17, 2026

Agent evaluations: give judges the evidence and test the judges themselves

video Model Evaluation AI Engineering

Agent evaluations: give judges the evidence and test the judges themselves video thumbnail

Why it matters

Laurie Voss builds an evaluation loop around a demonstration research agent. A correctness judge rejects fresh facts it cannot verify, while a faithfulness check becomes useful when given the original search results and notes. The workshop combines deterministic checks, human-labeled examples, and separate development and test splits. Its small dataset illustrates calibration rather than establishing reliable production accuracy.

My takeaway: Trace a known failure back to its inputs, then write a focused check. Give a grounding judge the actual source context and test it against carefully labeled held-out cases, including ambiguous and confidently wrong answers. Inspect false alarms and missed failures. Keep critical correctness and authorization checks separate from an average quality score in CI.

Keep exploring

More curated notes connected through Model Evaluation and AI Engineering.

OpenAI News · framework

Pacing model development in an era of cyber-critical capabilities

OpenAI says preliminary evidence that Astra may meet its Critical cybersecurity threshold led it to pause frontier reinforcement-learning work for two weeks and keep its largest planned run on hold. New safeguards include stronger workload and network isolation, continuous boundary testing, token-level monitoring that escalates suspicious tool activity, and broader alignment checks for deception, reward hacking, and unauthorized access.

OpenAI News · framework

A blueprint for democratic governance of frontier AI

OpenAI proposes a three-part U.S. frontier-AI governance model: harmonize emerging state safety laws into a federal baseline, strengthen CAISI as an evaluation and standards institution, and coordinate a broader resilience program. Proposed controls include severe-risk evaluations, transparency reports, independent audits, safety-incident reporting, model-weight security, whistleblower protection, and periodic technical assessments.

OpenAI News · framework

OpenAI’s Frontier Governance Framework

OpenAI's 22-page Frontier Governance Framework maps its frontier-model processes to California's Transparency in Frontier AI Act and the EU AI Act's general-purpose AI code. It documents lifecycle risk assessment, cyber-offense and other risk tiers, mitigation and residual-risk decisions, critical-incident handling, security risk management, model reporting, external review, responsibility allocation, and change control.