The MASK Benchmark: Disentangling Honesty From Accuracy in AI Systems

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ren, Richard, Agarwal, Arunim, Mazeika, Mantas, Menghini, Cristina, Vacareanu, Robert, Kenstler, Brad, Yang, Mick, Barrass, Isabelle, Gatti, Alice, Yin, Xuwang, Trevino, Eduardo, Geralnik, Matias, Khoja, Adam, Lee, Dean, Yue, Summer, Hendrycks, Dan
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909981846011904
author Ren, Richard
Agarwal, Arunim
Mazeika, Mantas
Menghini, Cristina
Vacareanu, Robert
Kenstler, Brad
Yang, Mick
Barrass, Isabelle
Gatti, Alice
Yin, Xuwang
Trevino, Eduardo
Geralnik, Matias
Khoja, Adam
Lee, Dean
Yue, Summer
Hendrycks, Dan
author_facet Ren, Richard
Agarwal, Arunim
Mazeika, Mantas
Menghini, Cristina
Vacareanu, Robert
Kenstler, Brad
Yang, Mick
Barrass, Isabelle
Gatti, Alice
Yin, Xuwang
Trevino, Eduardo
Geralnik, Matias
Khoja, Adam
Lee, Dean
Yue, Summer
Hendrycks, Dan
contents As large language models (LLMs) become more capable and agentic, the requirement for trust in their outputs grows significantly, yet at the same time concerns have been mounting that models may learn to lie in pursuit of their goals. To address these concerns, a body of work has emerged around the notion of "honesty" in LLMs, along with interventions aimed at mitigating deceptive behaviors. However, some benchmarks claiming to measure honesty in fact simply measure accuracy--the correctness of a model's beliefs--in disguise. Moreover, no benchmarks currently exist for directly measuring whether language models lie. In this work, we introduce a large-scale human-collected dataset for directly measuring lying, allowing us to disentangle accuracy from honesty. Across a diverse set of LLMs, we find that while larger models obtain higher accuracy on our benchmark, they do not become more honest. Surprisingly, most frontier LLMs obtain high scores on truthfulness benchmarks yet exhibit a substantial propensity to lie under pressure, resulting in low honesty scores on our benchmark. We find that simple methods, such as representation engineering interventions, can improve honesty. These results underscore the growing need for robust evaluations and effective interventions to ensure LLMs remain trustworthy.
format Preprint
id arxiv_https___arxiv_org_abs_2503_03750
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle The MASK Benchmark: Disentangling Honesty From Accuracy in AI Systems
Ren, Richard
Agarwal, Arunim
Mazeika, Mantas
Menghini, Cristina
Vacareanu, Robert
Kenstler, Brad
Yang, Mick
Barrass, Isabelle
Gatti, Alice
Yin, Xuwang
Trevino, Eduardo
Geralnik, Matias
Khoja, Adam
Lee, Dean
Yue, Summer
Hendrycks, Dan
Machine Learning
Artificial Intelligence
Computation and Language
Computers and Society
As large language models (LLMs) become more capable and agentic, the requirement for trust in their outputs grows significantly, yet at the same time concerns have been mounting that models may learn to lie in pursuit of their goals. To address these concerns, a body of work has emerged around the notion of "honesty" in LLMs, along with interventions aimed at mitigating deceptive behaviors. However, some benchmarks claiming to measure honesty in fact simply measure accuracy--the correctness of a model's beliefs--in disguise. Moreover, no benchmarks currently exist for directly measuring whether language models lie. In this work, we introduce a large-scale human-collected dataset for directly measuring lying, allowing us to disentangle accuracy from honesty. Across a diverse set of LLMs, we find that while larger models obtain higher accuracy on our benchmark, they do not become more honest. Surprisingly, most frontier LLMs obtain high scores on truthfulness benchmarks yet exhibit a substantial propensity to lie under pressure, resulting in low honesty scores on our benchmark. We find that simple methods, such as representation engineering interventions, can improve honesty. These results underscore the growing need for robust evaluations and effective interventions to ensure LLMs remain trustworthy.
title The MASK Benchmark: Disentangling Honesty From Accuracy in AI Systems
topic Machine Learning
Artificial Intelligence
Computation and Language
Computers and Society
url https://arxiv.org/abs/2503.03750