Auditing Language Models for Hidden Objectives

Auditing Language Models for Hidden Objectives

Why it matters

An auditing game in which blind teams look for a hidden objective planted in a model; 3 of 4 teams succeeded

Source — Primary sources first; secondary reports are labelled “Press”

Officialanthropic.comhttps://www.anthropic.com/research/auditing-hidden-objectives

Added in Week 35, 2026 · See that week

Same week

Figure announces BotQOfficialChina issues measures for labelling AI-generated and synthetic contentGovernmentCohere releases Command AOfficialJin et al. present Search-R1Paper

Nearby on the same axis

Monitoring Reasoning Models for MisbehaviorPaperOn the Biology of a Large Language ModelOfficialPreparedness Framework v2OfficialConstitutional ClassifiersOfficial