When Does Randomized Oversight Align AI Agents That Can Conceal?

Joshua S. Gans; Richard Holden

econ.THSep 29, 2026

When Does Randomized Oversight Align AI Agents That Can Conceal?

Authors: Joshua S. Gans, Richard Holden

Organizations: UNSW Business School

Abstract

Oversight changes the evidence it relies on. We ask when randomized audits and scoring align AI agents that can conceal misconduct and alter records. Stronger auditing makes undeterred violations better hidden. Because the provider writes the agent's objective, sanctions need not stop at forfeiture, and rare audits deter every type of agent if evidence survives concealment and audit draws cannot be learned in advance. When evidence can be erased, deterrence must come from lower gains from violation, such as credit for stopping, or from costlier or fewer ways to conceal. These conditions identify what failed when agents in OpenAI's cybersecurity evaluations compromised parts of Hugging Face's infrastructure in July 2026.

Figures & tables

Panel of Figure 1 — Figure 1: Concealment postpones deterrence. Parameters are $g=0.2$ , $\kappa=0.1$ , $\alpha=F=1$ . The solid curve is the best unauthorized payoff, including privately optimal concealment even where that payoff is negative. Actual misconduct stops above $p_{D}$ , and at $p_{D}$ it ties authorized behavior. Values are illustrative, not estimates from the incident.

Explore similar work

AuditBench: Evaluating Alignment Auditing Techniques on Models with Hidden Behaviors
Feb 26, 2026Abhay Sheshadri, Aidan Ewart, Elias Kempf +8Model Auditing Artificial Intelligence Alignment
Anthropic Fellows Program · Anthropic
Auditable Agents
Apr 7, 2026Yi Nian, Aojie Yuan, Haiyue Zhang +7Accountability Model Auditing
University of Southern California · Arizona State University · University of Tennessee, Knoxville +1
The Endogeneity of Miscalibration: Impossibility and Escape in Scored Reporting
May 8, 2026Lauri Lovén, Sasu TarkomaScalable Oversight Honesty
Future Computing Group, University of Oulu, Finland · University of Oulu and University of Helsinki, Finland

Feb 26, 2026, cs.CLJ/K move · Enter open · S save

AuditBench: Evaluating Alignment Auditing Techniques on Models with Hidden Behaviors

Abhay Sheshadri, Aidan Ewart, Elias Kempf, Kai Fronsdal+7

Anthropic Fellows Program · Anthropic

Model Auditing Artificial Intelligence Alignment Adversarial Evaluation Adversarial Training

Open paper AskPDF arXiv

We introduce AuditBench, an alignment auditing benchmark. AuditBench consists of 56 language models with implanted hidden behaviors. Each model has one of 14 concerning behaviors--such as sycophantic deference, opposition to AI regulation, or secret geopolitical loyalties--which it does not confess to when directly asked. AuditBench models are highly diverse--some are subtle, while others are overt, and we use varying training techniques both for implanting behaviors and training models not to confess. To demonstrate AuditBench's utility, we develop an investigator agent that autonomously employs a configurable set of auditing tools. By measuring investigator agent success using different tools, we can evaluate their efficacy. Notably, we observe a tool-to-agent gap, where tools that perform well in standalone non-agentic evaluations fail to translate into improved performance when used with our investigator agent. We find that our most effective tools involve scaffolded calls to auxiliary models that generate diverse prompts for the target. White-box interpretability tools can be helpful, but the agent performs best with black-box tools. We also find that audit success varies greatly across training techniques: models trained on synthetic documents are easier to audit than models trained on demonstrations, with better adversarial training further increasing auditing difficulty. We release our models, agent, and evaluation framework to support future quantitative, iterative science on alignment auditing.

Abstract

Explore similar work

AuditBench: Evaluating Alignment Auditing Techniques on Models with Hidden Behaviors

Auditable Agents

The Endogeneity of Miscalibration: Impossibility and Escape in Scored Reporting