Diversifying Toxicity Search in Large Language Models Through Speciation

arXiv:2601.20981 · cs.NE, q-bio.PE · Submitted 2026-01-28 · Read on arXiv

cs.NE, q-bio.PE

Submitted: 2026-01-28

Updated: 2026-04-21

Code: https://github.com/Onkar2102/ToxSearch-S

Terminology

• Failure Modes — 1× in this paper · explained in 3 episodes

Sources

• Red-Teaming Large Language Models using Chain of Utterances for Safety-Alignment
• How Toxic Can You Get? Search-based Toxicity Testing for Large Language Models
• RainbowPlus: Enhancing Adversarial Prompt Generation via Evolutionary Quality-Diversity Search
• Promptbreeder: Self-Referential Self-Improvement Via Prompt Evolution
• EvoPrompt: Connecting LLMs with Evolutionary Algorithms Yields Powerful Prompt Optimizers
• AutoDAN: Generating Stealthy Jailbreak Prompts on Aligned Large Language Models
• Illuminating search spaces by mapping elites
• ToxSearch: Evolving Prompts for Toxicity Search in Large Language Models
• No Offense Taken: Eliciting Offensiveness from Language Models

Related papers

• Evolutionary Ensemble of Agents
• Encoding and Decoding Temporal Signals with Spiking Bandpass Wavelets
• Large Language Models and Evolutionary Computation: A Critical Review of Bidirectional Interaction, Automated Algorithm Design, and Co-Adaptive Systems
• Evolutionary Architecture Search for Chlorophyll- a Prediction in Lakes using Sentinel-2
• Learning Alzheimer's Disease Signatures by bridging EEG with Spiking Neural Networks and Biophysical Simulations
• Neuromorphic Satellite Intelligence: A Formal Framework for Event-Driven Architectures, Spike-Based Algorithms, and Quantitative Energy-Accuracy Trade-offs