·
DataBubble
  • Home
  • Models
  • News
  • Compare
  • Boards
  • Pricing
  • About
  • Newsletter
  • Methodology
  • Contact
Latest
Iceland-based Treble raises $18 million for its voice simulation platform1h◆One Color Preprocessing Improves DSATUR2h◆Imitation Learning for Autonomous Driving in CARLA2h◆FairCompressAgent: An Agentic Framework for Fairness-Aware Model Compression for FPGA Deployment2h◆A Four-Stage Decomposition of Word-Problem Solving and Mechanistic Fragility in LLM Math Reasoning2h◆Learning Heterogeneous Preferences2h◆SNOMED CT Concept Recommendation from Masked Clinical Context2h◆The Inference Engineering Pareto Atlas: Which Optimizations Dominate the Cost, Quality, and Latency Frontier?2h◆ERPBench: A State-Grounded Evaluation Paradigm for Computer-Use Agents in Enterprise Software2h◆OBC-Prune: Outcome-Based Calibration for Large Reasoning Model Pruning2h◆Collaborative Memory for Multi-Agent VLM Systems2h◆Measuring AI Leadership: Development and Validation of a Multidimensional Measure for AI-Native Organizations2h◆Memory Has Geometry: Non-Uniform Geometric Memory for Long-Horizon Personalized AI2h◆When to Call an LLM: A Confidence-Gated Hybrid for Cost-Effective Emotion Recognition in Conversational AI2h◆Contiguity, Not Importance: Budgeted Repair of Stale KV Caches After Document Edits2h◆Missing Bridges: Composition-Aware Active Imitation Learning2h◆Anchoring What Matters: A Dual-Level Learning Framework for Visually-Grounded Multimodal Reasoning2h◆The Other Half of the Memory Wall: Serving 35B MoEs from SSD with Trained Routing Prediction2h◆Decodability is Not Causality: Dissociating Probe Readouts from Behavioral Drivers via SAE Decomposition2h◆When Is Graph Structure Worth Its Cost? The Case for Structure Pricing in Retrieval-Augmented Generation2h◆Iceland-based Treble raises $18 million for its voice simulation platform1h◆One Color Preprocessing Improves DSATUR2h◆Imitation Learning for Autonomous Driving in CARLA2h◆FairCompressAgent: An Agentic Framework for Fairness-Aware Model Compression for FPGA Deployment2h◆A Four-Stage Decomposition of Word-Problem Solving and Mechanistic Fragility in LLM Math Reasoning2h◆Learning Heterogeneous Preferences2h◆SNOMED CT Concept Recommendation from Masked Clinical Context2h◆The Inference Engineering Pareto Atlas: Which Optimizations Dominate the Cost, Quality, and Latency Frontier?2h◆ERPBench: A State-Grounded Evaluation Paradigm for Computer-Use Agents in Enterprise Software2h◆OBC-Prune: Outcome-Based Calibration for Large Reasoning Model Pruning2h◆Collaborative Memory for Multi-Agent VLM Systems2h◆Measuring AI Leadership: Development and Validation of a Multidimensional Measure for AI-Native Organizations2h◆Memory Has Geometry: Non-Uniform Geometric Memory for Long-Horizon Personalized AI2h◆When to Call an LLM: A Confidence-Gated Hybrid for Cost-Effective Emotion Recognition in Conversational AI2h◆Contiguity, Not Importance: Budgeted Repair of Stale KV Caches After Document Edits2h◆Missing Bridges: Composition-Aware Active Imitation Learning2h◆Anchoring What Matters: A Dual-Level Learning Framework for Visually-Grounded Multimodal Reasoning2h◆The Other Half of the Memory Wall: Serving 35B MoEs from SSD with Trained Routing Prediction2h◆Decodability is Not Causality: Dissociating Probe Readouts from Behavioral Drivers via SAE Decomposition2h◆When Is Graph Structure Worth Its Cost? The Case for Structure Pricing in Retrieval-Augmented Generation2h◆
News/Does Moral Reasoning Training Help or Hurt? Red-Teaming RL-Trained Ethical Agents with Persona Attacks
arxiv
PublishedSeptember 17, 2026 at 4:00 AM

Does Moral Reasoning Training Help or Hurt? Red-Teaming RL-Trained Ethical Agents with Persona Attacks

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

arXiv:2609.17552v1 Announce Type: new Abstract: Moral-reward RL can make language-model agents more cooperative, but whether that alignment survives adversarial persona pressure is unknown. Such attacks are realistic: retrieved context, tool outputs, or multi-turn framing can all inject role instruc

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 →

No replies yet. Be first.

Source
↗
arxiv
Read original ↗All from arxiv →

Related coverage

More from ARXIV
arxivOne Color Preprocessing Improves DSATUR2harxivImitation Learning for Autonomous Driving in CARLA2harxivFairCompressAgent: An Agentic Framework for Fairness-Aware Model Compression for FPGA Deployment2harxivA Four-Stage Decomposition of Word-Problem Solving and Mechanistic Fragility in LLM Math Reasoning2h
The Bubble Brief
WEEKLY

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

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

Originally published on arxiv ↗
HomeModelsNews