← Back to Model Beat
Research·4d ago·all news from November 18, 2026

Managing frontier AI risk: From training to deployment

OpenAI has published an overview of its safety protocols for frontier AI models, detailing procedures that range from initial training alignment to post-deployment monitoring. These measures include using isolated sandbox environments, blocking specific system outputs, and continuously tracking model behavior. By documenting these safeguards, the company aims to demonstrate how it attempts to mitigate potential risks and prevent misuse throughout the entire lifecycle of its most advanced systems.

Covered by 1 source

Related stories

ResearchOur first phase of embedded assessmentsNov 16 · 4 sourcesResearchResearch acceleration: The jagged frontierNov 17 · 4 sourcesResearchTop AI Models Regularly Identify Civilians as Military Targets in TestsNov 17 · 4 sourcesResearchCombatting adversarial distillationNov 20