Cover Image

Stochastic Monkeys at Play: Random Augmentations Cheaply Break LLM Safety Alignment

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Vega, Jason, Huang, Junsheng, Zhang, Gaokai, Kang, Hangoo, Zhang, Minjia, Singh, Gagandeep
Format:	Preprint
Published:	2024
Subjects:	Machine Learning Artificial Intelligence
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

TRAP: Targeted Redirecting of Agentic Preferences
by: Kang, Hangoo, et al.
Published: (2025)

Bypassing the Safety Training of Open-Source LLMs with Priming Attacks
by: Vega, Jason, et al.
Published: (2023)

Matching Ranks Over Probability Yields Truly Deep Safety Alignment
by: Vega, Jason, et al.
Published: (2025)

TriPlay-RL: Tri-Role Self-Play Reinforcement Learning for LLM Safety Alignment
by: Tan, Zhewen, et al.
Published: (2026)

SynCode: LLM Generation with Grammar Augmentation
by: Ugare, Shubham, et al.
Published: (2024)

The Geometry of Alignment Collapse: When Fine-Tuning Breaks Safety
by: Springer, Max, et al.
Published: (2026)

Learning a Pessimistic Reward Model in RLHF
by: Xu, Yinglun, et al.
Published: (2025)

Data Shifts Hurt CoT: A Theoretical Study
by: Yin, Lang, et al.
Published: (2025)

Probabilistic Trust Intervals for Out of Distribution Detection
by: Singh, Gagandeep, et al.
Published: (2021)

PuzzleMoE: Efficient Compression of Large Mixture-of-Experts Models via Sparse Expert Merging and Bit-packed inference
by: Zhao, Yushu, et al.
Published: (2025)

The Attacker in the Mirror: Breaking Self-Consistency in Safety via Anchored Bipolicy Self-Play
by: La Malfa, Gabriele, et al.
Published: (2026)

Better LLM Reasoning via Dual-Play
by: Zhang, Zhengxin, et al.
Published: (2025)

LASA: Language-Agnostic Semantic Alignment at the Semantic Bottleneck for LLM Safety
by: Yang, Junxiao, et al.
Published: (2026)

VoltanaLLM: Feedback-Driven Frequency Control and State-Space Routing for Energy-Efficient LLM Serving
by: Yu, Jiahuan, et al.
Published: (2025)

Any-Depth Alignment: Unlocking Innate Safety Alignment of LLMs to Any-Depth
by: Zhang, Jiawei, et al.
Published: (2025)

LLM Safety Alignment is Divergence Estimation in Disguise
by: Haldar, Rajdeep, et al.
Published: (2025)

Alignment and Safety in Large Language Models: Safety Mechanisms, Training Paradigms, and Emerging Challenges
by: Lu, Haoran, et al.
Published: (2025)

How Catastrophic is Your LLM? Certifying Risk in Conversation
by: Wang, Chengxiao, et al.
Published: (2025)

Universal Black-Box Reward Poisoning Attack against Offline Reinforcement Learning
by: Xu, Yinglun, et al.
Published: (2024)

Differentiated Directional Intervention A Framework for Evading LLM Safety Alignment
by: Zhang, Peng, et al.
Published: (2025)

Augmented Lagrangian Multiplier Network for State-wise Safety in Reinforcement Learning
by: Zhang, Jiaming, et al.
Published: (2026)

Understand What LLM Needs: Dual Preference Alignment for Retrieval-Augmented Generation
by: Dong, Guanting, et al.
Published: (2024)

Alignment Dynamics in LLM Fine-Tuning
by: Huang, Yuhan, et al.
Published: (2026)

Reward-Augmented Data Enhances Direct Preference Alignment of LLMs
by: Zhang, Shenao, et al.
Published: (2024)

SafeTuneBed: A Toolkit for Benchmarking LLM Safety Alignment in Fine-Tuning
by: Hossain, Saad, et al.
Published: (2025)

Uncovering Logit Suppression Vulnerabilities in LLM Safety Alignment
by: Li, Yuxi, et al.
Published: (2024)

SuperInfer: SLO-Aware Rotary Scheduling and Memory Management for LLM Inference on Superchips
by: Yu, Jiahuan, et al.
Published: (2026)

Symmetry-Breaking Augmentations for Ad Hoc Teamwork
by: Hammond, Ravi, et al.
Published: (2024)

Why Do Some Inputs Break Low-Bit LLM Quantization?
by: Chang, Ting-Yun, et al.
Published: (2025)

Curriculum Learning for Safety Alignment
by: Kumar, Sandeep, et al.
Published: (2026)

RSPO: Regularized Self-Play Alignment of Large Language Models
by: Tang, Xiaohang, et al.
Published: (2025)

Light Alignment Improves LLM Safety via Model Self-Reflection with a Single Neuron
by: Shen, Sicheng, et al.
Published: (2026)

The Causal Impact of Tool Affordance on Safety Alignment in LLM Agents
by: Yu, Shasha, et al.
Published: (2026)

Attention Eclipse: Manipulating Attention to Bypass LLM Safety-Alignment
by: Zaree, Pedram, et al.
Published: (2025)

Certifying Knowledge Comprehension in LLMs
by: Chaudhary, Isha, et al.
Published: (2024)

Formal Synthesis of Certifiably Robust Neural Lyapunov-Barrier Certificates
by: Wang, Chengxiao, et al.
Published: (2026)

The Blessing and Curse of Dimensionality in Safety Alignment
by: Teo, Rachel S. Y., et al.
Published: (2025)

From Parameter Dynamics to Risk Scoring : Quantifying Sample-Level Safety Degradation in LLM Fine-tuning
by: Wang, Xiao, et al.
Published: (2026)

Breaking the Safety-Capability Tradeoff: Reinforcement Learning with Verifiable Rewards Maintains Safety Guardrails in LLMs
by: Cho, Dongkyu Derek, et al.
Published: (2025)

Large Language Monkeys: Scaling Inference Compute with Repeated Sampling
by: Brown, Bradley, et al.
Published: (2024)