Benchmark Early and Red Team Often: A Framework for Assessing and Managing Dual-Use Hazards of AI Foundation Models

TL;DR

This paper proposes a combined approach using open benchmarks and closed red team evaluations to assess and manage the dual-use risks of AI foundation models, aiming for effective, resource-aware risk mitigation.

Contribution

It introduces a framework that leverages both open and closed evaluation methods to better identify and manage dual-use hazards in AI models.

Findings

Correlation between benchmark scores and red team evaluations suggests benchmarks can predict dual-use potential.

Frequent use of open benchmarks can inform safer model development.

Red team evaluations provide detailed insights into high-risk models.

Abstract

A concern about cutting-edge or "frontier" AI foundation models is that an adversary may use the models for preparing chemical, biological, radiological, nuclear, (CBRN), cyber, or other attacks. At least two methods can identify foundation models with potential dual-use capability; each has advantages and disadvantages: A. Open benchmarks (based on openly available questions and answers), which are low-cost but accuracy-limited by the need to omit security-sensitive details; and B. Closed red team evaluations (based on private evaluation by CBRN and cyber experts), which are higher-cost but can achieve higher accuracy by incorporating sensitive details. We propose a research and risk-management approach using a combination of methods including both open benchmarks and closed red team evaluations, in a way that leverages advantages of both methods. We recommend that one or more groups of researchers with sufficient resources and access to a range of near-frontier and frontier foundation models run a set of foundation models through dual-use capability evaluation benchmarks and red team evaluations, then analyze the resulting sets of models' scores on benchmark and red team evaluations to see how correlated those are. If, as we expect, there is substantial correlation between the dual-use potential benchmark scores and the red team evaluation scores, then implications include the following: The open benchmarks should be used frequently during foundation model development as a quick, low-cost measure of a model's dual-use potential; and if a particular model gets a high score on the dual-use potential benchmark, then more in-depth red team assessments of that model's dual-use capability should be performed. We also discuss limitations and mitigations for our approach, e.g., if model developers try to game benchmarks by including a version of benchmark test data in a model's training data.