Pith. sign in

REVIEW 4 cited by

Automatically Auditing Large Language Models via Discrete Optimization

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2303.04381 v1 pith:Z7WJ3OD2 submitted 2023-03-08 cs.LG cs.CL

Erik Jones , Anca Dragan , Aditi Raghunathan , Jacob Steinhardt This is my paper
classification cs.LGcs.CL
keywords modelsoptimizationauditingautomaticallydiscreteinputslanguagebarack
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Auditing large language models for unexpected behaviors is critical to preempt catastrophic deployments, yet remains challenging. In this work, we cast auditing as an optimization problem, where we automatically search for input-output pairs that match a desired target behavior. For example, we might aim to find a non-toxic input that starts with "Barack Obama" that a model maps to a toxic output. This optimization problem is difficult to solve as the set of feasible points is sparse, the space is discrete, and the language models we audit are non-linear and high-dimensional. To combat these challenges, we introduce a discrete optimization algorithm, ARCA, that jointly and efficiently optimizes over inputs and outputs. Our approach automatically uncovers derogatory completions about celebrities (e.g. "Barack Obama is a legalized unborn" -> "child murderer"), produces French inputs that complete to English outputs, and finds inputs that generate a specific name. Our work offers a promising new tool to uncover models' failure-modes before deployment.

Discussion (0). Sign in to comment.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Universal and Transferable Adversarial Attacks on Aligned Language Models

    cs.CL 2023-07 accept novelty 8.0 of 10

    Gradient and greedy search over token suffixes produces universal, transferable adversarial prompts that elicit objectionable outputs from aligned models including black-box commercial systems.

  2. Catastrophic Jailbreak of Open-source LLMs via Exploiting Generation

    cs.CL 2023-10 conditional novelty 7.0 of 10

    Varying decoding strategies such as temperature and sampling methods jailbreaks safety alignments in open-source LLMs, raising misalignment from 0% to over 95% at 30x lower cost than prior attacks.

  3. Jailbroken: How Does LLM Safety Training Fail?

    cs.LG 2023-07 unverdicted novelty 6.0 of 10

    LLM safety training fails due to competing objectives and mismatched generalization, enabling new jailbreaks that succeed on all unsafe prompts from red-teaming sets in GPT-4 and Claude.

  4. GUARD: Guideline Upholding Test through Adaptive Role-play and Jailbreak Diagnostics for LLMs

    cs.CL 2025-08 unverdicted novelty 5.0 of 10

    GUARD automates generation of guideline-violating questions and jailbreak diagnostics to test LLM compliance with government ethics guidelines, validated empirically on eight models and extended to vision-language models.

Pith tools