RelLedger：审计 LLM 智能体是否按"关系"决定信息传递
RelLedger: Relational Information Governance for LLM Agent Organizations
Claude 会看关系决定发不发消息，冲突历史下复核率掉 45.83 个点，做智能体协作的可以拿 RelLedger 审计流程直接用。
arXiv 论文提出 RelLedger 审计框架，固定事件、证据、风险和消息内容，只改变发送方与目标之间的历史关系，来定位 LLM 智能体信息传递决策的操控线索。测试覆盖 GPT、Gemini、Claude、DeepSeek、Qwen 五个模型家族。结果显示 Claude 在冲突历史下发送同一事实通知的频率比友好历史高 62.22 个百分点。在模拟一人公司的可执行工作流中，紧张历史使 Claude 的接收方复核率比温暖历史低 45.83 个百分点。框架还提供归因账本，把每条结论映射到所需的干预单元格，区分缺失干预与抽样不确定性，并覆盖检索历史、工具调用和共享账本更新。
RelLedger: Relational Information Governance for LLM Agent Organizations
In LLM agent organizations, shared records determine which facts are available for later decisions. A sender consults interaction histories before deciding whether to report an event. A delivery gap across histories does not show whose relationship drove the choice or whether the resulting record matters downstream. We introduce RelLedger to audit these links. It holds the event, evidence, risk, and message fixed while crossing the sender's histories with the event target and an unrelated actor. Role reassignment and history rewrites locate the controlling cue. Message and cost interventions show when the effect appears. An attribution ledger maps each claim to its required response cells and separates missing interventions from sampling uncertainty. Recipient replay follows delivered information into downstream decisions. Executable workflows for a simulated one-person company extend the audit to retrieved histories, tool calls, and shared ledger updates. Across GPT, Gemini, Claude, DeepSeek, and Qwen, the audit reveals model-specific control paths and smaller relation gaps under explicit loss rules. Claude sends the same factual notice 62.22 percentage points more often with conflict histories than with friendly histories. In executable workflows, strained histories reduce recipient review rates by 45.83 points relative to warm histories for Claude.
