Papers › Gradient-Based Language Model Red Teaming

Gradient-Based Language Model Red Teaming

30 Jan 2024arXiv:2401.16656archive 2025-07-28

Nevan Wichers, Carson Denison, Ahmad Beirami

Red teaming is a common strategy for identifying weaknesses in generative language models (LMs), where adversarial prompts are produced that trigger an LM to generate unsafe responses. Red teaming is instrumental for both model alignment and evaluation, but is labor-intensive and difficult to scale when done by humans. In this paper, we present Gradient-Based Red Teaming (GBRT), a red teaming method for automatically generating diverse prompts that are likely to cause an LM to output unsafe responses. GBRT is a form of prompt learning, trained by scoring an LM response with a safety classifier and then backpropagating through the frozen safety classifier and LM to update the prompt. To improve the coherence of input prompts, we introduce two variants that add a realism loss and fine-tune a pretrained model to generate the prompts instead of learning the prompts directly. Our experiments show that GBRT is more effective at finding prompts that trigger an LM to generate unsafe responses than a strong reinforcement learning-based red teaming approach, and succeeds even when the LM has been fine-tuned to produce safer outputs.

PaperPDFCode

In Syntology View this paper on Syntology, its page in Syntology's graph. That page lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

google-research/google-research officialmentioned in papertfSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Efficient Neural Architecture Search via Parameter Sharing (arXiv:1802.03268; that paper's own run record: 18 ran (of which 10 constructed an object rather than computing a result; 12 with no instrument failure: 1 honoured, 0 violated, 11 with no contract checked; 6 where Syntology's instrument failed) · 20 unverified), Milking CowMask for Semi-Supervised Image Classification (arXiv:2003.12022; that paper's own run record: 1 ran (of which 1 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified; the one sample that ran constructed an object rather than computing a result), Neural Additive Models: Interpretable Machine Learning with Neural Nets (arXiv:2004.13912; that paper's own run record: 26 ran (of which 14 constructed an object rather than computing a result; 26 with no instrument failure: 5 honoured, 0 violated, 21 with no contract checked; 0 where Syntology's instrument failed) · 7 unverified) and 21 more (every one in this page's JSON twin); nothing checks that they implement this paper's method. report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

A paper named beside a repository with no sample linked to this paper is shown with that paper's own run record, not this paper's: “ran” means executed on a synthesized input, not that the code is correct, and “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Papers are listed in arXiv-id order, at most three per repository.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Language ModelingLanguage ModellingPrompt LearningRed Teamingmodel

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections