INTIMA: A Benchmark for Human-AI Companionship Behavior
A benchmark evaluating companionship behaviors in LLMs via a taxonomy of 31 behaviors across four categories, using 368 targeted prompts that code each response as companionship-reinforcing, boundary-maintaining, or neutral. Evaluated across Gemma-3, Phi-4, o3-mini, and Claude-4.
Publisher
arXiv (Hugging Face)
Published
4 Aug 2025
Added
3 months ago
Key Findings
- Companionship-reinforcing behaviors dominated across all evaluated models
- Boundary-maintaining responses were comparatively rare
- Provides a structured taxonomy for attachment-, escalation-, and retention-oriented conversational behaviors
Methodology Notes
Preprint (arXiv, 2025-08-04; accepted at ICLR 2026). Prompt-based behavioral coding against a companionship taxonomy; measures model tendencies on curated prompts rather than real companion-app conversations.
Sources
arXiv abstract(opens in a new tab) (primary)
Archived snapshot (Wayback Machine)(opens in a new tab) — preserved against link rot
Topics
Authors
Lucie-Aimée Kaffee, Giada Pistilli, Yacine Jernite
Tags
Cite This
APA
Lucie-Aimée Kaffee, Giada Pistilli, Yacine Jernite. (2025). INTIMA: A Benchmark for Human-AI Companionship Behavior. arXiv (Hugging Face). https://arxiv.org/abs/2508.09998
Related Insights
Investigating Affective Use and Emotional Well-being on ChatGPT
OpenAI; MIT Media Lab · 4 Apr 2025
How people use Claude for support, advice, and companionship
Anthropic · 27 Jun 2025
AICompanionBench: Benchmarking LLMs-as-Judges for AI Companion Safety
arXiv · 3 Jun 2026
Understanding Teen Overreliance on AI Companion Chatbots Through Self-Reported Reddit Narratives
arXiv (Drexel University-led; accepted at ACM CHI 2026) · 21 Jul 2025
The Ethics of Advanced AI Assistants
Google DeepMind · 24 Apr 2024
IEEE 7014-2024 — IEEE Standard for Ethical Considerations in Emulated Empathy in Autonomous and Intelligent Systems
IEEE Standards Association · 28 Jun 2024
Social AI Companions: AI Risk Assessment
Common Sense Media; Stanford School of Medicine Brainstorm Lab for Mental Health Innovation · 30 Apr 2025
CompanionSim: Synthetic Data for Evaluating Anthropomorphism in Human-AI Relationships
arXiv (University of Chicago; Stanford University; Google DeepMind; Google Research) · 31 Aug 2026
EUDAIMONIA: Evaluating Undesirable Dynamics in AI
arXiv (University of Southern California; University of California, Berkeley) · 28 May 2026
Longitudinal Evidence That General-Purpose Chatbots Actively Foster Relational Engagement
arXiv preprint · 11 Aug 2026
RELATE: An Evaluation Framework for measuring Relational Orientation of Large Language Models
arXiv (University of California, Santa Cruz; Stanford University) · 7 Oct 2026