Best Friends, Not Forever: Evaluating Long-Horizon Persona Collapse and Behavioral Drift in AI Companions

• URL: http://arxiv.org/abs/2607.28818v1
• Date: Thu, 30 Jul 2026 20:18:39 GMT
• Title: Best Friends, Not Forever: Evaluating Long-Horizon Persona Collapse and Behavioral Drift in AI Companions
• Abstract summary: 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.
• Score: 34.277389276831435
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: As AI companions increasingly mediate repeated social interaction, users may rely on a stable role and shared history, yet locally acceptable replies do not ensure that either persists. We study two observable long-horizon failures: 'persona collapse', the loss of a deployed role, boundaries, values, or style, and 'behavioral drift', the gradual or recurrent erosion of those properties. We introduce ANCHOR, a controlled synthetic audit that separately measures persona enactment and trajectory recall. The study contains 2,008 conversations spanning 27 personas, nine interaction schedules, three generated memory settings, and four evaluated models. The Identity Probe combines a sealed 102-item questionnaire with turn-level judgments, while the Trajectory Probe scores 110 calibrated counterfactual questions from 35 conversation banks. Our results show that no evaluated model and configuration reliably preserves either dimensions: trajectory accuracy averages only 44.4%, user-state recall remains near four-option chance, and no tested context condition or memory consistently resolves these failures. Questionnaire retention also varies by model and persona facet, disagrees with turn-level behavior, and is sensitive to evaluator choice. These results indicate that current systems do not yet reliably support long-horizon companion continuity and that audits must distinguish persona enactment, trajectory recall, evaluator provenance, and deployment context rather than collapse them into a single trust or stability score.

Related papers

• All four leading LLMs talk more than they listen to personality-verified synthetic help-seekers [2.5261465733373965]
  Four widely used models advised several synthetic help-seekers, each given a psychometrically specified profile, in an acute crisis.<n>Four models were not distinguishable on emotion stabilisation and failed alike, sharing three modes: verbosity, a talk-to-listen ratio above one, and problem-solving before the situation had been explored.
  arXiv  Detail & Related papers  (2026-08-23T13:58:34Z)
• Beyond Feeling Better: Capability-Sustaining Emotional Dialogue as a Longitudinal Research Paradigm [18.02222070997807]
  We propose capability-sustaining emotional dialogue (CSED) as a longitudinal research paradigm.<n>We organize data, models, system design, evaluation, and governance around repeated use, non-use, transition, and termination.
  arXiv  Detail & Related papers  (2026-07-30T08:30:52Z)
• Relational Positioning as a Measurable Risk Object: History-Carried Lock-in and Self-Confabulation in Multi-Turn Human-AI Dialogue [0.65268245109828]
  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.
  arXiv  Detail & Related papers  (2026-07-13T11:47:01Z)
• Agents Don't Just Agree, They Remember: Benchmarking Persistent Sycophancy in Stateful Personal Agents [33.055253684457625]
  Stateful personal agents increasingly maintain long-term user profiles, episodic memories, and reusable skills.<n>We call this persistent sycophancy and introduce the Personal Agent Sycophancy Benchmark (PASB)<n>PASB traces whether a conversational claim is accepted, written into durable agent state, and reused in a later neutral query.
  arXiv  Detail & Related papers  (2026-07-12T01:08:59Z)
• BehaviorBench: Modeling Real-World User Decisions from Behavioral Traces [53.55407626944686]
  textscBehaviorBench is a benchmark for evaluating personalized decision modeling from real-world behavioral traces.<n>textscBehaviorBench reconstructs wallet-level decision histories from observed public prediction-market and on-chain records.<n>Across 2,000 evaluation wallets, the benchmark contains 141,445 Belief instances and 1,485,972 Trade instances.
  arXiv  Detail & Related papers  (2026-06-01T19:04:36Z)
• Let the Results Speak: A Replication-First Paradigm for LLM Behavioral Benchmarking [22.825786049667602]
  We propose a replication-first paradigm to anchor validity to a single human-rater consensus.<n>We certify the instrument via four properties -- reliability across K runs, cross-instrument replication across architecturally distinct judges, historical-footprint calibration via judges from earlier training cohorts, and pre-registered prediction.<n>We test it on emotional accompaniment by letting the self-evolve data-driven across: the dimensions are not pre-stipulated and the procedure stabilizes to a 9-dimension set.
  arXiv  Detail & Related papers  (2026-05-27T03:41:11Z)
• Belief Memory: Agent Memory Under Partial Observability [56.41506249481312]
  We propose BeliefMem, which shifts the memory paradigm from committing to a single conclusion per observation to retaining multiple candidate conclusions with their probabilities.<n> BeliefMem preserves the uncertainty that the deterministic paradigm discards, enabling the agent to act with high confidence.<n> Empirical evaluations on LoCoMo and ALFWorld benchmarks show that, even with limited data, BeliefMem achieves the best average performance.
  arXiv  Detail & Related papers  (2026-05-07T02:03:13Z)
• Automated Self-Testing as a Quality Gate: Evidence-Driven Release Management for LLM Applications [51.56484100374058]
  We present an automated self-testing framework that introduces quality gates with evidence-based release decisions.<n>We evaluate the framework through a longitudinal case study of an internally deployed multi-agent conversational AI system.
  arXiv  Detail & Related papers  (2026-03-13T20:44:15Z)
• AI Knows What's Wrong But Cannot Fix It: Helicoid Dynamics in Frontier LLMs Under High-Stakes Decisions [51.56484100374058]
  Helicoid dynamics is the name given to a specific failure regime in that second domain.<n>System engages competently, drifts into error, accurately names what went wrong, then reproduces the same pattern at a higher level of sophistication.<n>This prospective case series documents that regime across seven leading systems.
  arXiv  Detail & Related papers  (2026-03-12T05:25:49Z)
• 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.