SovereignNegotiation-Bench: Evaluating User-Owned Personal Agents In Delegated Bargaining Under Privacy, Consent, Evidence, And Institutional Pressure

arXiv:2607.02814 · cs.MA, cs.AI, cs.CY · Submitted 2026-07-02 · Read on arXiv

cs.MA, cs.AI, cs.CY

Submitted: 2026-07-02

Updated: 2026-10-01

Code: https://github.com/microsoft/social-reasoning-bench

Terminology

• Benchmark — 3× in this paper · explained in 2 episodes
• Grounding — 2× in this paper · explained in 3 episodes
• Trace — 1× in this paper · explained in 2 episodes

Sources

• Cooperation, Competition, and Maliciousness: LLM-Stakeholders Interactive Negotiation
• Firewalls to Secure Dynamic LLM Agentic Networks
• AI loyalty: A New Paradigm for Aligning Stakeholder Interests
• Magentic Marketplace: An Open-Source Environment for Studying Agentic Markets
• $\tau^2$-Bench: Evaluating Conversational Agents in a Dual-Control Environment
• How Well Can LLMs Negotiate? NegotiationArena Platform and Analysis
• CI-Bench: Benchmarking Contextual Integrity of AI Assistants on Synthetic Data
• AgentDojo: A Dynamic Environment to Evaluate Prompt Injection Attacks and Defenses for LLM Agents
• Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection
• MAGPIE: A benchmark for Multi-AGent contextual PrIvacy Evaluation
• Governing AI Agents
• Deal or No Deal? End-to-End Learning for Negotiation Dialogues
• Whose Side Is Your Agent On? Multi-Party Principal Loyalty in LLM Agents
• Used Car Salesbots? Honesty and Credulity of LLMs as Bargaining Agents under Partial Information
• gpt-oss-120b & gpt-oss-20b Model Card
• Towards Understanding Sycophancy in Language Models
• Confuse the Model, Control the Flow: Understanding and Mitigating Privacy Leakage from LLM Agents with Information Flow Control
• AgentLeak: A Benchmark for Internal-Channel Privacy Leakage in Multi-Agent LLM Systems
• TERMS-Bench: Diagnosing LLM Negotiation Agents Beyond Deal Rate
• POLAR-Bench: A Diagnostic Benchmark for Privacy-Utility Trade-offs in LLM Agents

Related papers

• Highway Congestion Reduction through Reinforcement Learning Based Eulerian Headway Control
• ORACLE-SWE: Quantifying the Contribution of Critical Information Signals on SWE Agents
• You Only Align Once: Propagating Cooperative Behaviors in Multi-Agent Systems through Seed Agents
• Deny Without Disabling: Authorization-Paired Evaluation and Control for Multi-Agent Systems
• MA-SAPO: Multi-Agent Reasoning for Score-Aware Prompt Optimization
• PeroMAS: A Multi-agent System of Perovskite Material Discovery