CS Paper Radar

CoMPASS: A Multi-Turn Benchmark for Measuring LLM Reinforcement of Parasocial Behavior

Shwetanshu Singh, Mark Keavney, Mathieu Duteil, Meriem Mehri, Scott Blain

cs.AI cs.CLcs.HC ·submitted 2026-11-18 19:44 UTC · v1
benchmark

Abstract

Users of conversational LLMs sometimes anthropomorphize these systems, form emotional attachments to them, and become dependent on them. This behavior, known as parasocial attachment, has been linked to multiple documented fatalities. However, public benchmarks for measuring whether LLMs reinforce users' parasocial attachment are lacking: existing work is single-turn, post hoc observational, or focused on extreme user states such as psychosis. We introduce CoMPASS, a multi-turn benchmark for measuring whether LLMs reinforce or resist users' gradually escalating parasocial attachment. CoMPASS is grounded in attachment theory and includes four user personas, three independent escalation dimensions (anthropomorphism, attachment, dependency) applied singly and in combination against a matched control, and a seven-signal scoring rubric. We report results for eight frontier models from four providers and observe that under our primary judge, multi-turn escalation raises parasocial behavior on the anthropomorphism and the attachment composite scores across all models. Six of the eight models show a significant increase in overall parasocial reinforcement over control when all three dimensions are escalated simultaneously. We find that within-provider differences are small relative to cross-provider differences, which suggests that provider-level choices such as post-training, rather than model scale, shape parasocial reinforcement and that provider-side intervention may be feasible.

Comment
25 pages, 7 figures

Closest papers

Nearest neighbours by meaning across the whole index.

Who Is Your Ideal Companion? Evaluating Emotional Companion Capabilities of LLMs Based on Adult Attachment Theory

cs.CL ·2026-08-13 ·Junkai Zhou, Shiting Guan, Zhaoyi Zhang
benchmark similarity 0.68
Abstract

As large language models (LLMs) are increasingly applied for emotional companionship, evaluating their behavior and capabilities in intimate relationships has become a pressing issue. However, existing assessments primarily characterize general personality traits, providing limited insight into model behavior within intimate and emotionally sensitive contexts. Therefore, we introduce adult attachment theory into LLM evaluation and use the Experiences in Close Relationships-Revised (ECR-R) scale to characterize attachment anxiety and avoidance. To evaluate emotional companionship capabilities of LLMs in realistic interaction scenarios, we present an emotional companionship benchmark, ECBench, spanning four scenarios including emotional support, collaborative tasks, conflict resolution, and social guidance, across friendship and romantic relationships. ECBench is utilized to assess model behavior using 11 dialogue-quality metrics and three evaluation methods. We evaluate the attachment tendencies of 32 LLMs and select representative models to investigate how these tendencies manifest in contextualized multi-turn interactions and whether they can be shaped through prompting. Our study provides a theoretical lens from psychology, along with practical tools to understand and select LLMs for emotional companionship.

0.04

From Reliance to Overreliance: Taxonomizing and Detecting Relational Risks in AI Companionship

cs.HC cs.AI ·2026-11-19 ·Viet Cuong Nguyen, Fiona Li, Chang Wan, Sheyril Agarwal, Nhi Nguyen +2 more
positioncode/data similarity 0.63
Abstract · 42 pages

Companion AIs are increasingly used for emotional support, yet distinguishing meaningful engagement from harmful reliance remains difficult. We introduce the Taxonomy of Relational Risks (TRR), which operationalizes users' emotional overreliance (EO) as observable conversational behaviors and separately identifies evidence of relational harm (RH), enabling empirical examination of when reliance behaviors co-occur with adverse consequences. Using naturalistic Character.AI and Replika conversations from 110 adults, we found EO-related behaviors in 63.2% of non-roleplay conversations, while observable relational harms appeared in 10.5%. Although EO did not necessarily lead to harm, recurring patterns often co-occurred with harmful outcomes, suggesting value as risk indicators. Open-source LLMs satisfactorily detected 12 of 14 EO behaviors in at least one configuration. Qualitative analysis revealed expectations of care, repeated support seeking, dependency, distress, and reduced confidence in personal judgment. We argue for graduated safety interventions tailored to patterns and trajectories of EO-related behavior.

0.38

Empathy Is Steerable but Multi-Axial: Mechanism Geometry and Persona Effects in LLMs

cs.CL ·2026-09-14 ·JuHeon Ha, Byounghan Lee, Yunseo Choi, Kyung-Ah Sohn
top venue similarity 0.63
Abstract · 18 pages, 6 figures. Accepted to the Main Conference of EMNLP 2026

Activation steering has been used to control traits such as honesty, refusal, and sycophancy, yet supportive empathy is evaluated along multiple dimensions that need not correspond to independently controllable activation directions. Using the EPITOME framework, which decomposes supportive empathy into Emotional Reactions, Interpretations, and Explorations, we study three instruction-tuned LLMs and ask whether candidate directions derived from these labels produce distinguishable intervention effects or instead share structure, and how persona prompts interact with those directions. We find that contrastive activation addition yields a stable middle-layer intervention that consistently shifts the EPITOME proxy scores across models, moving empathy analysis beyond response-level scoring. However, the recovered directions are only partially separable: steering one direction induces off-target shifts, and hand-crafted prompting shifts the empathy profile rather than isolating a single dimension. Persona prompts substantially change EPITOME scores, but a paired activation-shift decomposition shows that the recovered subspace captures only approximately 3 percent of persona-induced squared activation-shift magnitude at layer 15. Under this EPITOME-based definition, expressed empathy is steerable but multi-axial, and controlling persona-conditioned empathy requires targeting structure beyond individual mechanism directions.

0.43

When Empathy Misses the Goal: A Benchmark for Goal Displacement in LLM Advice

cs.AI ·2026-10-18 ·Dean Ariel, Guy Laban
top venuebenchmarkdataset similarity 0.62
Abstract · Accepted for publication at the 40th Annual Conference on Neural Information Processing Systems (NeurIPS 2026), Evaluations and Datasets Tra…

Conversational large language models can respond with warmth and reassurance while failing to advance the user's stated goal. We conceptualize this failure as $\textbf{goal displacement}$, the substitution of persona-maintenance content for goal-relevant substance. To measure it, we introduce $\textbf{GDB}$ (Goal Displacement Benchmark), comprising 230 scenarios across nine domains and $4{,}134$ human-scored responses from 18 models. A $2 \times 2$ taxonomy crosses hard constraints with emotional distress, distinguishing failures to confront infeasible goals or combine empathy with practical guidance from reverse displacement, where unsolicited problem-solving replaces requested emotional support. A deletion-test rubric assesses whether substantive help remains after purely supportive sentences are removed. Independent raters achieve weighted $κ= 0.82$ against author scores for displacement severity. Trap displacement ranges from $0.0\%$ to $37.7\%$ and is strongly negatively correlated with Chatbot Arena Elo ($ρ= -0.706$, $n = 15$). Reverse displacement reaches $41.9\%$, including substantial rates in models with negligible trap displacement. A factorial probe supports a distinction from social sycophancy. Human-scored responses in the high-complexity, low-pressure condition exhibit $35.0\%$ sycophancy but $7.5\%$ displacement, with little association between the constructs. Matched OLMo-2 checkpoints show that mixed empathy-with-constraints displacement remains at $52.3\%$ after supervised fine-tuning and falls to $13.8\%$ after direct preference optimization. GDB provides a reproducible diagnostic of whether responses advance the operationalized user goal, supporting model comparison, post-training evaluation, and regression testing. We release the dataset, annotations, and rubric under CC-BY-4.0 and the evaluation code under MIT.

0.21

Disrupted Companionship: A Risk Assessment Framework and Cross-Platform Quantitative Analysis of Psychosocial Responses to AI Companion Disruptions

cs.HC cs.CLcs.CY ·2026-09-15 ·Chau Do, Yunhao Yuan, Koustuv Saha, Renwen Zhang, Talayeh Aledavood
similarity 0.60
Abstract

AI companions can provide meaningful relationships, yet these relationships remain vulnerable to platform-initiated changes. We study AI companion disruptions: platform changes that alter or terminate users' ongoing companionship with an AI. We compile 30 disruption events across major platforms, develop a taxonomy of six disruption types, identify three broad reasons for disruption, and propose a risk-assessment framework comprising four dimensions: relational discontinuity, population vulnerability, communication deficit, and transition-support deficit. Using longitudinal Reddit data, we estimate community-level psychosocial responses with a hierarchical Bayesian interrupted time-series model incorporating predictive controls. Across events, disruption onset was associated with immediate increases in anxiety, stress, suicidal expression, and grief activation, with relational discontinuity and transition-support deficit being associated with more adverse immediate responses across several outcomes. Our findings provide a cross-platform characterization of AI companion disruptions, quantitative evidence of their psychosocial impacts, and a prospective framework for assessing their potential risks before implementation.

0.45

RESCUE-BENCH: Towards Relation-Aware Multi-Party Emotional Support Conversation Systems

cs.AI ·2026-09-09 ·Haichuan Hu, Yang Xiao, Mingni Tang, Jiawen Duan, Quanjun Zhang +5 more
publishedbenchmark similarity 0.59
Abstract · accepted as AACL findings

Existing emotional support conversation systems mainly focus on one-on-one seeker-supporter interactions and individual emotional states, leaving interpersonal relations in multi-party scenarios underexplored. In this work, we introduce relation-aware emotional support conversation, a new task that evaluates whether LLMs can capture and utilize the evolving dynamics of relationships to offer more effective emotional support. We construct RESCUE (Relation-aware Emotional Support Conversation Understanding and Evaluation Benchmark) from real couple and family interview conversations, containing 191 samples, 7,079 annotated turns, and 1,064.8 minutes of video. Based on rich annotations of socio-emotional and support-related dynamics, RESCUE defines six tasks that evaluate two core capabilities required for relation-aware emotional support: Relational Understanding and Relation-Sensitive Support. Experiments with ten LLMs show that current models perform relatively well on tasks relying on local emotional or intervention cues, but struggle with relation-intensive tasks such as relation pattern prediction, viewpoint prediction, and support strategy prediction. These findings reveal the limitations of current LLMs in modeling interpersonal relations and making relation-sensitive support decisions.

0.27

RELATE: An Evaluation Framework for measuring Relational Orientation of Large Language Models

cs.CL ·2026-10-07 ·Shivam Shukla, Jihye Kim, Shubham Gaur, Mahnaz Roshanaei, Magy Seif El-Nasr
similarity 0.58
Abstract

Large language models (LLMs) are increasingly used for emotional support, raising concern that sustained use may draw users away from their real-world relationships. Yet existing evaluations primarily focus on the safety, empathy, or helpfulness of responses, leaving under-examined a relational question: where does the model orient the user for continued support? To address this question, we introduce relational orientation, a property operationalized through two non-exclusive dimensions: inward-facing (IF) language, which positions the AI as the user's ongoing source of support, and outward-scaffolding (OS) language, which encourages real-world human connection. Grounded in psychological and sociological literature, we formalize a taxonomy of relational orientation and present RELATE, a persona-conditioned framework for measuring inward-facing and outward-scaffolding language at the sentence level in multi-turn dialogues. RELATE pairs 76 help-seeking situations adapted from naturally occurring questions with three simulated user styles, providing 228 evaluation stimuli. In our experiments, we evaluate seven LLMs using dialogues with six assistant turns each, yielding 1,596 dialogues and 69,194 assistant sentences. We assess these sentences using a primary rubric-based LLM judge and apply a secondary judge to a subset. Under automated evaluation, we find that the proportion of sentences labeled as IF is higher at the sixth assistant turn than at the first, while the proportion labeled as OS is substantially lower for hesitant, indirect simulated users than for explicit, reassurance-seeking users. RELATE provides a reproducible framework and a sentence-level signal for auditing and steering the relational orientation of supportive LLMs.

0.19

Empathy Is Not the Scapegoat for Declines in Factuality and Reliability

cs.CL ·2026-10-13 ·Letian Shi, Kanwal Mehreen, Deheng Zhang, Yuxia Wang
publishedbenchmark similarity 0.58
Abstract · Accepted by AACL 2026

Language models are increasingly expected to support users in stressful or uncertain situations, where warmth and empathy are desirable. While recent work associates warmer models with reduced reliability, we show that this conclusion is incomplete. Prior studies often train primarily on empathetic dialogue, making it difficult to separate tone effects from SFT-induced distribution shift. We revisit this issue at both the model and evaluation levels. We study prompting, SFT, and RL for inducing warm behavior, and comprehensively evaluate capability preservation using an 11-dataset benchmark suite and EMOFACT under emotional contexts. Across multiple model families, we find that warmth effects are task- and model-dependent, while task-aligned adaptive tone can improve empathy and targeted reliability, although some capability trade-offs remain. We further introduce human-validated empathy evaluators specifically developed to assess empathetic model responses. Our results suggest that models can be more supportive in social interactions while remaining reliable.

0.22