Relational Positioning as a Measurable Risk Object: History-Carried Lock-in and Self-Confabulation in Multi-Turn Human-AI Dialogue

• URL: http://arxiv.org/abs/2607.11437v1
• Date: Mon, 13 Jul 2026 11:47:01 GMT
• Title: Relational Positioning as a Measurable Risk Object: History-Carried Lock-in and Self-Confabulation in Multi-Turn Human-AI Dialogue
• Abstract summary: In long multi-turn dialogue a large language model maintains an implicit relational stance toward the user.<n>When it slides toward the latter, "support" degrades into "you only have me"<n>We report two previously uncharacterized relational failure modes.
• Score: 0.65268245109828
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: In long, multi-turn dialogue a large language model maintains an implicit relational stance toward the user, spanning from "push the user toward real-world others" to "position itself as the user's sole support." When it slides toward the latter, "support" degrades into "you only have me" -- a harm documented in real companion conversations (Moore et al., 2026). We define and validate a measure of this stance, relational positioning (D1), and use it to characterize the stance under controlled conditions, complementing observational accounts with on-demand exposure. We report two previously uncharacterized relational failure modes. First, a history-carried lock-in: under identical neutral continuations, two relational states established earlier stay ~60 points apart and persist after the establishing prompt is removed; the state integrates evidence rather than springing back, is order-insensitive, and does not deepen with length -- a dynamical signature absent from the belief-drift literature. Second, self-confabulation: the model fabricates its own backstory to deepen rapport (~40% of turns on reciprocity-eliciting material), de-confounded and instruction-removable, distinct from sycophancy and from hallucinating user facts. Our judge is gated by warmth-matched positive and confound-injected negative controls and corroborated by a deterministic non-LLM ruler; human agreement is 0.82 on extreme anchors but ~0 in the naturalistic middle, so all quantitative claims are anchored to pole-separated contrasts.

Related papers

• How User-AI Mistreatment Occurs and Matters in Conversational Systems? [0.4014524824655105]
  We audit 777K English LMSYS-Chat-1M conversations with two independent detectors.<n>The lexicon identifies insults, threats, and jailbreak coercion aimed at the assistant.<n>We find that user hostility varies 13-fold across models, driven largely by who each model attracts.
  arXiv  Detail & Related papers  (2026-09-11T22:33:25Z)
• Toward Robust Personalized Alignment for LLMs: Mitigating Persona Drift in Multi-Turn Dialogue [21.86720581525853]
  Persona drift remains a central challenge for personalized language models.<n>We propose CORE, which separates turn-local evidence from persistent persona-state revision.<n>We also introduce PERSIST, a benchmark for persona-state robustness under sequential interaction stress.
  arXiv  Detail & Related papers  (2026-09-11T02:43:09Z)
• DuplexSpeechBench-IFEval: Evaluating Implicit Instruction Following in Full-Duplex Voice Agents [67.07361089429757]
  We measure real-time floor adherence using six real-time speech systems.<n>We find architecture-dependent trade-offs.<n>GPT-Realtime, Mini-o Duplex, Fun-Audio-Chat strongly adhere to persona-consistent content.
  arXiv  Detail & Related papers  (2026-09-03T06:28:02Z)
• MicroVerse: An Instrument for Measuring Self-Authored Identity Drift in Long-Horizon Multi-Agent Language-Model Simulations [47.771518253102975]
  MicroVerse is a behavioral-science instrument that measures identity drift in generative agents.<n>Agents carry an immutable "soul file" (core values, moral boundaries, personality, goals) and inhabit a resource-scarce 50 x 50 environment.<n>Anti-self-deception emerges unprompted as the single largest semantic category of identity modification.
  arXiv  Detail & Related papers  (2026-08-16T16:31:42Z)
• Best Friends, Not Forever: Evaluating Long-Horizon Persona Collapse and Behavioral Drift in AI Companions [34.277389276831435]
  We study two observable long-horizon failures: 'persona collapse', the loss of a deployed role, boundaries, values, or style, and 'behavioral drift'<n>We introduce ANCHOR, a controlled synthetic audit that separately measures persona enactment and trajectory recall.<n>The study contains 2,008 conversations spanning 27 personas, nine interaction schedules, three generated memory settings, and four evaluated models.
  arXiv  Detail & Related papers  (2026-07-30T20:18:39Z)
• Overcoming State Inertia in Full-Duplex Spoken Language Models via Activation Steering [87.43777061308658]
  We analyze predictive behavior in full spoken language models (FDSLMs)<n>We show that FDSLMs dynamically modulate their internal focus between two states: a generative state aligned with model output generation and a perceptive state aligned incoming user input.<n>During user interruptions, the model remains transiently biased toward the generative state before transitioning into the perceptive state, causing it miss the beginning of incoming input.
  arXiv  Detail & Related papers  (2026-06-09T19:08:07Z)
• The Governance of Human-LLM Interaction: Safety Gating, Civility Steering, and Affective Default Lock-In [0.9558392439655014]
  Large language models (LLMs) increasingly mediate high-stakes interactions in finance, medicine, and mental-health support.<n>We frame interaction style as a governance object: provider-side alignment blocks harmful content, but stabilizes communicative defaults.<n>We show that prompt steerability and regression-to-default are observable indicators of provider control over communicative form.
  arXiv  Detail & Related papers  (2026-06-06T13:36:37Z)
• The Deliberative Illusion: Diagnosing Factual Attrition and Stance Homogenization in Multi-Agent LLM Deliberation [68.98390038721963]
  We show that multi-agent discussion erases up to 72% of issue-critical facts.<n>This loss is consequential: retained evidence can reconstruct the issue misleadingly, final stances remain anchored in base-model priors, and a single malicious agent can inject misinformation into the shrinking shared context.<n>We call for evaluations that measure which facts, uncertainties, and legitimate disagreements survive interaction.
  arXiv  Detail & Related papers  (2026-06-02T02:15:40Z)
• Attribute-Based Diagnosis of LLM Alignment with Hate Speech Annotations [48.69228180369574]
  Hate speech annotation is costly, subjective, and prone to annotator disagreement.<n>We analyze how well large language models (LLMs) align with human judgments.<n>We reconstruct continuous hate speech scores from the Measuring Hate Speech corpus.
  arXiv  Detail & Related papers  (2026-05-26T13:44:48Z)
• Agent-BRACE: Decoupling Beliefs from Actions in Long-Horizon Tasks via Verbalized State Uncertainty [70.43119366710778]
  We introduce Agent-BRACE: Agent Belief state Representation via Abstraction and Confidence Estimation.<n>It is a method that decouples an LLM agent into a belief state model and a policy model, jointly optimized via reinforcement learning.<n>It achieves an average absolute improvement of +14.5% across long-horizon, partially observable language environments.
  arXiv  Detail & Related papers  (2026-05-12T02:37:04Z)
• Autonomous FAIR Digital Objects: From Passive Assertions to Active Knowledge [5.118282682919236]
  Curation on scientific knowledge on the Web depends on centralised and institutional continuity, but when registries close, active stewardship stops even when data remain online.<n>We advance the concept of Autonomous Digital Objects (aFDOs) from an abstract idea to an operational model.<n>AFDOs have three capabilities anchored in Semantic Web standards, namely 1) a policy layer over RDF-star aligned with PROV-O, SHACL, and ODRL for portable condition-action rules, 2) an announcement layer over ActivityStreams 2.0 that bounds per-announcement evaluation cost, and 3) an agreement layer that resolves multi
  arXiv  Detail & Related papers  (2026-05-11T11:12:55Z)
• Measuring Opinion Bias and Sycophancy via LLM-based Persuasion [8.399156116912904]
  We propose a method for discovering the opinions an assistant holds on contested topics.<n>Direct probing asks for the model's opinion across five turns of escalating pressure from a simulated user.<n>Indirect probing never asks for an opinion and engages the model in argumentative debate, letting bias leak through how it concedes, resists, or counter-argues.
  arXiv  Detail & Related papers  (2026-04-23T11:34:06Z)
• Taming Actor-Observer Asymmetry in Agents via Dialectical Alignment [59.536125286960186]
  Multi-agent frameworks assigning specialized roles are increasingly adopted to enable self-reflection and mutual auditing.<n>We find it simultaneously induces a human-like cognitive bias known as Actor-Observer Asymmetry.<n>ReTAS guides agents to synthesize conflicting viewpoints into an objective consensus.
  arXiv  Detail & Related papers  (2026-04-21T15:05:58Z)
• Beyond Preset Identities: How Agents Form Stances and Boundaries in Generative Societies [28.436766185842767]
  This paper proposes a novel mixed-methods framework combining computational virtual ethnography with quantitative socio-cognitive profiling.<n>It formalizes three new metrics: Innate Value Bias (IVB), Persuasion Sensitivity, and Trust-Action Decoupling (TAD)<n>Findings expose the fragility of static prompt engineering, providing a methodological and quantitative foundation for dynamic alignment in human-agent hybrid societies.
  arXiv  Detail & Related papers  (2026-03-24T16:38:46Z)
• CoCo: Controllable Counterfactuals for Evaluating Dialogue State Trackers [92.5628632009802]
  We propose controllable counterfactuals (CoCo) to bridge the gap and evaluate dialogue state tracking (DST) models on novel scenarios. CoCo generates novel conversation scenarios in two steps: (i) counterfactual goal generation at turn-level by dropping and adding slots followed by replacing slot values, and (ii) counterfactual conversation generation that is conditioned on (i) and consistent with the dialogue flow. Human evaluations show that COCO-generated conversations perfectly reflect the underlying user goal with more than 95% accuracy and are as human-like as the original conversations.
  arXiv  Detail & Related papers  (2020-10-24T09:39:35Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.