Cover Image

The Alignment Floor: How Persona Customization Breaks Safety in Weakly-Aligned LLMs

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Zhang, Xing, Wang, Guanghui, Cui, Yanwei, Qiu, Wei, Li, Ziyuan, Zhu, Bing, He, Peiyang
Format:	Preprint
Published:	2026
Subjects:	Human-Computer Interaction Artificial Intelligence Computation and Language
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

CloChat: Understanding How People Customize, Interact, and Experience Personas in Large Language Models
by: Ha, Juhye, et al.
Published: (2024)

Personas with Attitudes: Controlling LLMs for Diverse Data Annotation
by: Fröhling, Leon, et al.
Published: (2024)

Direct Advantage Regression: Aligning LLMs with Online AI Reward
by: He, Li, et al.
Published: (2025)

OPeRA: A Dataset of Observation, Persona, Rationale, and Action for Evaluating LLMs on Human Online Shopping Behavior Simulation
by: Wang, Ziyi, et al.
Published: (2025)

Using Contextually Aligned Online Reviews to Measure LLMs' Performance Disparities Across Language Varieties
by: Tang, Zixin, et al.
Published: (2025)

Prompting ChatGPT for Translation: A Comparative Analysis of Translation Brief and Persona Prompts
by: He, Sui
Published: (2024)

Aligning LLMs with Individual Preferences via Interaction
by: Wu, Shujin, et al.
Published: (2024)

Guardrails Beat Guidance: A Large-Scale Study of Rules, Skills, and Persistent Configuration for Coding Agents
by: Zhang, Xing, et al.
Published: (2026)

Ratchet: A Minimal Hygiene Recipe for Self-Evolving LLM Agents
by: Zhang, Xing, et al.
Published: (2026)

Prompt Optimization Is a Coin Flip: Diagnosing When It Helps in Compound AI Systems
by: Zhang, Xing, et al.
Published: (2026)

CURE: Cultural Understanding and Reasoning Evaluation - A Framework for "Thick" Culture Alignment Evaluation in LLMs
by: Vo, Truong, et al.
Published: (2025)

Dialogue Language Model with Large-Scale Persona Data Engineering
by: Hong, Mengze, et al.
Published: (2024)

Do LLMs Make Mistakes Like Students? Exploring Natural Alignment between Language Models and Human Error Patterns
by: Liu, Naiming, et al.
Published: (2025)

Aligning Language Models with Demonstrated Feedback
by: Shaikh, Omar, et al.
Published: (2024)

Creating and Evaluating Personas Using Generative AI: A Scoping Review of 81 Articles
by: Amin, Danial, et al.
Published: (2025)

CALYPSO: LLMs as Dungeon Masters' Assistants
by: Zhu, Andrew, et al.
Published: (2023)

LLMs as Workers in Human-Computational Algorithms? Replicating Crowdsourcing Pipelines with LLMs
by: Wu, Tongshuang, et al.
Published: (2023)

Library Drift: Diagnosing and Fixing a Silent Failure Mode in Self-Evolving LLM Skill Libraries
by: Zhang, Xing, et al.
Published: (2026)

Experience Compression Spectrum: Unifying Memory, Skills, and Rules in LLM Agents
by: Zhang, Xing, et al.
Published: (2026)

PersoPilot: An Adaptive AI-Copilot for Transparent Contextualized Persona Classification and Personalized Response Generation
by: Afzoon, Saleh, et al.
Published: (2026)

PersoDPO: Scalable Preference Optimization for Instruction-Adherent, Persona-Grounded Dialogue via Multi-LLM Evaluation
by: Afzoon, Saleh, et al.
Published: (2026)

ReSpark: Leveraging Previous Data Reports as References to Generate New Reports with LLMs
by: Tian, Yuan, et al.
Published: (2025)

Generating Educational Materials with Different Levels of Readability using LLMs
by: Huang, Chieh-Yang, et al.
Published: (2024)

CUPID: Evaluating Personalized and Contextualized Alignment of LLMs from Interactions
by: Kim, Tae Soo, et al.
Published: (2025)

BEYOND DIALOGUE: A Profile-Dialogue Alignment Framework Towards General Role-Playing Language Model
by: Yu, Yeyong, et al.
Published: (2024)

SensorPersona: An LLM-Empowered System for Continual Persona Extraction from Longitudinal Mobile Sensor Streams
by: Yang, Bufang, et al.
Published: (2026)

Customizing ChatGPT for Second Language Speaking Practice: Genuine Support or Just a Marketing Gimmick?
by: Meng, Fanfei
Published: (2026)

Mind the Value-Action Gap: Do LLMs Act in Alignment with Their Values?
by: Shen, Hua, et al.
Published: (2025)

Effects of Theory of Mind and Prosocial Beliefs on Steering Human-Aligned Behaviors of LLMs in Ultimatum Games
by: Yadav, Neemesh, et al.
Published: (2025)

Aligning Stuttered-Speech Research with End-User Needs: Scoping Review, Survey, and Guidelines
by: Toyin, Hawau Olamide, et al.
Published: (2026)

Aligning Human-AI-Interaction Trust for Mental Health Support: Survey and Position for Multi-Stakeholders
by: Sun, Xin, et al.
Published: (2026)

LLM Agent Meets Agentic AI: Can LLM Agents Simulate Customers to Evaluate Agentic-AI-based Shopping Assistants?
by: Sun, Lu, et al.
Published: (2025)

ValueCompass: A Framework for Measuring Contextual Value Alignment Between Human and LLMs
by: Shen, Hua, et al.
Published: (2024)

LLMs' ways of seeing User Personas
by: Panda, Swaroop
Published: (2024)

Comparing How a Chatbot References User Utterances from Previous Chatting Sessions: An Investigation of Users' Privacy Concerns and Perceptions
by: Cox, Samuel Rhys, et al.
Published: (2023)

Beyond Prompts: Learning from Human Communication for Enhanced AI Intent Alignment
by: Kim, Yoonsu, et al.
Published: (2024)

Towards Stable and Personalised Profiles for Lexical Alignment in Spoken Human-Agent Dialogue
by: Schaaij, Keara, et al.
Published: (2025)

Completing A Systematic Review in Hours instead of Months with Interactive AI Agents
by: Qiu, Rui, et al.
Published: (2025)

After Talking with 1,000 Personas: Learning Preference-Aligned Proactive Assistants From Large-Scale Persona Interactions
by: Xuan, Ziyi, et al.
Published: (2026)

Robots in the Middle: Evaluating LLMs in Dispute Resolution
by: Tan, Jinzhe, et al.
Published: (2024)