·
DataBubble
  • Home
  • Models
  • News
  • Compare
  • Boards
  • Pricing
  • About
  • Newsletter
  • Methodology
  • Contact
Latest
Bringing Value Models Back: Generative Critics for Value Modeling in LLM Reinforcement Learning6h◆Subagents vs Agent Skills: Executing Reusable Knowledge for Long-Horizon Agentic Tasks6h◆Distribution-Consistent Inference for Dynamic Sparse Mixture-of-Experts6h◆In RAG We Trust? Measuring Robustness of Retrieval-Augmented Generation Under Document Poisoning6h◆FrontierChallenge: Evaluating Scientific Workflow Completion6h◆IBIB: A Protocol for Measuring Enterprise AI Systems by Serving Route, Not Model Identifier6h◆AgenticGen: Reward-Guided Agentic Video Generation for Advertising6h◆Strangers to Themselves: What Language Models Say About Themselves Is Generic6h◆Omni Interaction Agent Technical Report6h◆CoGReV: A Confidence-Gated Post-Hoc Non-Monotonic Belief Revision Framework for Phishing Website Classification6h◆Bit-Flip Attacks on Vision-Language-Action Models: Action-Decoding Architecture Shapes the Vulnerability6h◆Can Foundation Models Moderate Online Content? Evaluating Instruction- vs. Example-Driven Policy Operationalization6h◆RiLM: Parameter-Efficient Language Modeling via Geodesic Decoding6h◆Tracing Computation Density in LLMs6h◆Cultural Binding Heads in Language Models6h◆Direct Diversity Optimization for Diverse Successful Trajectories in Preference Post-Training6h◆Left-Branching Transformers Excel at Right-Branching Languages: Data Shapes Word Order Preferences in Language Models6h◆Palmyra x6 Technical Report: An Agentic, Tool-Use Model Post-Trained via Anchored Supervised Fine-Tuning6h◆'Ghaib in Translation' aka Unseen Harm: Measuring Cross-Script Safety Inconsistency with 'Missed-in-Urdu' Scores in LLM Hate Speech Detection6h◆DexterSQL: Deep Schema Exploration and Rule-based Correction for Text-to-SQL Generation6h◆Bringing Value Models Back: Generative Critics for Value Modeling in LLM Reinforcement Learning6h◆Subagents vs Agent Skills: Executing Reusable Knowledge for Long-Horizon Agentic Tasks6h◆Distribution-Consistent Inference for Dynamic Sparse Mixture-of-Experts6h◆In RAG We Trust? Measuring Robustness of Retrieval-Augmented Generation Under Document Poisoning6h◆FrontierChallenge: Evaluating Scientific Workflow Completion6h◆IBIB: A Protocol for Measuring Enterprise AI Systems by Serving Route, Not Model Identifier6h◆AgenticGen: Reward-Guided Agentic Video Generation for Advertising6h◆Strangers to Themselves: What Language Models Say About Themselves Is Generic6h◆Omni Interaction Agent Technical Report6h◆CoGReV: A Confidence-Gated Post-Hoc Non-Monotonic Belief Revision Framework for Phishing Website Classification6h◆Bit-Flip Attacks on Vision-Language-Action Models: Action-Decoding Architecture Shapes the Vulnerability6h◆Can Foundation Models Moderate Online Content? Evaluating Instruction- vs. Example-Driven Policy Operationalization6h◆RiLM: Parameter-Efficient Language Modeling via Geodesic Decoding6h◆Tracing Computation Density in LLMs6h◆Cultural Binding Heads in Language Models6h◆Direct Diversity Optimization for Diverse Successful Trajectories in Preference Post-Training6h◆Left-Branching Transformers Excel at Right-Branching Languages: Data Shapes Word Order Preferences in Language Models6h◆Palmyra x6 Technical Report: An Agentic, Tool-Use Model Post-Trained via Anchored Supervised Fine-Tuning6h◆'Ghaib in Translation' aka Unseen Harm: Measuring Cross-Script Safety Inconsistency with 'Missed-in-Urdu' Scores in LLM Hate Speech Detection6h◆DexterSQL: Deep Schema Exploration and Rule-based Correction for Text-to-SQL Generation6h◆
News/Low-Agreeableness Persona Conditioning for Safe LLM Fine-Tuning
arxiv
PublishedJune 29, 2026 at 4:00 AM
—neutral

Low-Agreeableness Persona Conditioning for Safe LLM Fine-Tuning

Source
arxiv.orgfull article ↗
Read on arxiv→
Publisher summary· verbatim

arXiv:2606.27709v1 Announce Type: cross Abstract: Recent work has shown that fine-tuning large language models (LLMs) for social warmth degrades factual reliability and increases sycophancy. We investigate a related but distinct failure mode: warmth fine-tuning also weakens adversarial safety, makin

Stay posted· Newsletter

A 5-min weekly brief — top movers, price watch, story of the week.

// no spam · unsubscribe one-click · free forever

Discussion
Source
↗
arxiv
Read original ↗All from arxiv →
Tags
03
#safety#fine-tuning#language-models

No replies yet. Be first.

Source
↗
arxiv
Read original ↗All from arxiv →
Tags
03
#safety#fine-tuning#language-models

Related coverage

More from ARXIV
arxivBringing Value Models Back: Generative Critics for Value Modeling in LLM Reinforcement Learning6harxivSubagents vs Agent Skills: Executing Reusable Knowledge for Long-Horizon Agentic Tasks6harxivDistribution-Consistent Inference for Dynamic Sparse Mixture-of-Experts6harxivIn RAG We Trust? Measuring Robustness of Retrieval-Augmented Generation Under Document Poisoning6h
The Bubble Brief
WEEKLY

Read safety insights every Tuesday — top movers, new releases, story of the week.

// no spam · unsubscribe one-click · free forever

Originally published on arxiv ↗
HomeModelsNews