[2507.21509] Persona Vectors: Monitoring and Controlling Character Traits in Language Models

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs. arXiv Operational Status Get status notifications via email or slack

[2507.21509] Persona Vectors: Monitoring and Controlling Character Traits in Language Models Skip to main content arXiv is now an independent nonprofit! Learn more × Search arXiv Press Enter to search · Advanced search --> Computer Science > Computation and Language arXiv:2507.21509 (cs) [Submitted on 29 Jul 2025 ( v1 ), last revised 5 Sep 2025 (this version, v3)] Title: Persona Vectors: Monitoring and Controlling Character Traits in Language Models Authors: Runjin Chen , Andy Arditi , Henry Sleight , Owain Evans , Jack Lindsey View a PDF of the paper titled Persona Vectors: Monit

related reading

• [2507.21509] Persona Vectors: Monitoring and Controlling Character Traits in Language Modelsarxiv.org
• [2401.05566] Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Trainingarxiv.org
• [2406.11717] Refusal in Language Models Is Mediated by a Single Directionarxiv.org
• [2309.01809] Are Emergent Abilities in Large Language Models just In-Context Learning?arxiv.org
• Persona vectors: Monitoring and controlling character traits in language models \ Anthropicanthropic.com
• [2503.10965] Auditing language models for hidden objectivesarxiv.org
• [2601.10387] The Assistant Axis: Situating and Stabilizing the Default Persona of Language Modelsarxiv.org
• [2512.19428] Attention Is Not What You Needarxiv.org
• [2510.07686] Stress-Testing Model Specs Reveals Character Differences among Language Modelsarxiv.org
• [2506.00751] Alignment Revisited: Are Large Language Models Consistent in Stated and Revealed Preferences?arxiv.org
• [2312.15159] Understanding the Potential of FPGA-Based Spatial Acceleration for Large Language Model Inferencearxiv.org
• [2407.15549] Latent Adversarial Training Improves Robustness to Persistent Harmful Behaviors in LLMsarxiv.org