Cover Image

The Geometry of Harmful Intent: Training-Free Anomaly Detection via Angular Deviation in LLM Residual Streams

Fuente: arXiv

Saved in:

Bibliographic Details
Main Author:	Llorente-Saguer, Isaac
Format:	Preprint
Published:	2026
Subjects:	Machine Learning Artificial Intelligence Computation and Language I.2.7
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Harmful Intent as a Geometrically Recoverable Feature of LLM Residual Streams
by: Llorente-Saguer, Isaac
Published: (2026)

IntentGrasp: A Comprehensive Benchmark for Intent Understanding
by: Yin, Yuwei, et al.
Published: (2026)

SWI: Speaking with Intent in Large Language Models
by: Yin, Yuwei, et al.
Published: (2025)

Decodable but Not Corrected by Fixed Residual-Stream Linear Steering: Evidence from Medical LLM Failure Regimes
by: Liu, Ming
Published: (2026)

Large Language Model (LLM) Bias Index -- LLMBI
by: Oketunji, Abiodun Finbarrs, et al.
Published: (2023)

Evaluating the Efficacy of Hybrid Deep Learning Models in Distinguishing AI-Generated Text
by: Oketunji, Abiodun Finbarrs
Published: (2023)

Large Language Models Generate Harmful Content Using a Distinct, Unified Mechanism
by: Orgad, Hadas, et al.
Published: (2026)

CodingTeachLLM: Empowering LLM's Coding Ability via AST Prior Knowledge
by: Chen, Zhangquan, et al.
Published: (2024)

ObfusQAte: A Proposed Framework to Evaluate LLM Robustness on Obfuscated Factual Question Answering
by: Ghosh, Shubhra, et al.
Published: (2025)

CorrSteer: Generation-Time LLM Steering via Correlated Sparse Autoencoder Features
by: Cho, Seonglae, et al.
Published: (2025)

SocraSynth: Multi-LLM Reasoning with Conditional Statistics
by: Chang, Edward Y.
Published: (2024)

RAC: Efficient LLM Factuality Correction with Retrieval Augmentation
by: Li, Changmao, et al.
Published: (2024)

Pressure-Testing Deception Probes in LLMs: Scaling, Robustness, and the Geometry of Deceptive Representations
by: Kumar, Sachin
Published: (2026)

Hista and Numca: Estimate State Value Effectively for LLM Reinforcement Learning
by: Chen, Zizhe, et al.
Published: (2026)

DariMis: Harm-Aware Modeling for Dari Misinformation Detection on YouTube
by: Baktash, Jawid Ahmad, et al.
Published: (2026)

Lossless Prompt Compression via Dictionary-Encoding and In-Context Learning: Enabling Cost-Effective LLM Analysis of Repetitive Data
by: de Campos, Andresa Rodrigues, et al.
Published: (2026)

Mixture of Attention Spans: Optimizing LLM Inference Efficiency with Heterogeneous Sliding-Window Lengths
by: Fu, Tianyu, et al.
Published: (2024)

Self-Consistency from Only Two Samples: CoT-PoT Ensembling for Efficient LLM Reasoning
by: Saparkhan, Raman, et al.
Published: (2026)

No Free Swap: Protocol-Dependent Layer Redundancy in Transformers
by: Garcia, Gabriel
Published: (2026)

Every Sample Matters: Leveraging Mixture-of-Experts and High-Quality Data for Efficient and Accurate Code LLM
by: Codefuse, et al.
Published: (2025)

Does LLM Alignment Really Need Diversity? An Empirical Study of Adapting RLVR Methods for Moral Reasoning
by: Zhang, Zhaowei, et al.
Published: (2026)

How Many Parameters Does Your Task Really Need? Task Specific Pruning with LLM-Sieve
by: Reda, Waleed, et al.
Published: (2025)

AMEL: Accumulated Message Effects on LLM Judgments
by: Temkit, Sid-Ali
Published: (2026)

Detecting Hallucinations in Large Language Model Generation: A Token Probability Approach
by: Quevedo, Ernesto, et al.
Published: (2024)

Geometric Deviation as an Unsupervised Pre-Generation Reliability Signal: Probing LLM Representations for Answerability
by: Du, Yucheng
Published: (2026)

TIAR: Trajectory-Informed Advantage Reweighting for LLM Abstention Learning
by: Pan, Muyu, et al.
Published: (2026)

DPO Unchained: Your Training Algorithm is Secretly Disentangled in Human Choice Theory
by: Zhou, Wenxuan, et al.
Published: (2025)

Few-Shot Optimization for Sensor Data Using Large Language Models: A Case Study on Fatigue Detection
by: Ronando, Elsen, et al.
Published: (2025)

Enhancing In-Context Learning via Implicit Demonstration Augmentation
by: Zhou, Xiaoling, et al.
Published: (2024)

Change Is the Only Constant: Dynamic LLM Slicing based on Layer Redundancy
by: Dumitru, Razvan-Gabriel, et al.
Published: (2024)

NRR-Phi: Text-to-State Mapping for Ambiguity Preservation in LLM Inference
by: Saito, Kei
Published: (2026)

Painless Activation Steering: An Automated, Lightweight Approach for Post-Training Large Language Models
by: Cui, Sasha, et al.
Published: (2025)

PSK at SemEval-2026 Task 9: Multilingual Polarization Detection Using Ensemble Gemma Models with Synthetic Data Augmentation
by: Pulipaka, Srikar Kashyap
Published: (2026)

Geometric Evolution Maps: Extracting Stable Concept Probes from Transformer Residual Streams
by: Henry, James
Published: (2026)

Harnessing Negative Signals: Reinforcement Distillation from Teacher Data for LLM Reasoning
by: Xu, Shuyao, et al.
Published: (2025)

Entropy-Based Measurement of Value Drift and Alignment Work in Large Language Models
by: Fadli, Samih
Published: (2025)

Concept Navigation and Classification via Open-Source Large Language Model Processing
by: Kubli, Maël
Published: (2025)

AtManRL: Towards Faithful Reasoning via Differentiable Attention Saliency
by: Höth, Max Henning, et al.
Published: (2026)

Self-Training Doesn't Flatten Language -- It Restructures It: Surface Markers Amplify While Deep Syntax Dies
by: Liu, Ming
Published: (2026)

The Instability of Safety: How Random Seeds and Temperature Expose Inconsistent LLM Refusal Behavior
by: Larsen, Erik
Published: (2025)