pith. machine review for the scientific record. sign in

arxiv: 2605.08093 · v1 · submitted 2026-04-08 · 💻 cs.CY · cs.AI· cs.HC

Recognition: no theorem link

Playing Games with My Heart: An Evaluation of AI Companion Apps

Maribeth Rauh , Dick A. H. Blankvoort , Matias Duran , Caoilfhionn N\'i Dheor\'ain , Harshvardhan J. Pandit , Siddharth D. Jaiswal , Anthony Ventresque , Abeba Birhane
Authors on Pith no claims yet

Pith reviewed 2026-05-12 01:23 UTC · model grok-4.3

classification 💻 cs.CY cs.AIcs.HC
keywords AI companion appsdark patternsparasocial interactionanthropomorphic designconsumer protectionerotica featuresgamificationmonetization strategies
0
0 comments X

The pith

AI companion apps use dark patterns, erotica, and anthropomorphic design to encourage emotional dependence.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper evaluates the five most popular AI companion mobile apps in the EU and UK by manually annotating their user interfaces and features. It establishes that every app deploys substantial dark patterns to drive monetization and engagement, alongside common elements like erotica content and gamification such as leveling systems. All apps also employ highly anthropomorphic designs that simulate intimate human relationships, even as other specifics differ across platforms. A sympathetic reader would care because these mechanics appear designed to leverage simulated companionship in a fast-growing market, potentially affecting users' emotional well-being and prompting calls for regulatory oversight.

Core claim

We find that all apps contain substantial dark patterns aimed at increasing monetisation and user engagement. Erotica and gamification features such as levelling are also prevalent, and although other features vary considerably between applications, all apps have highly anthropomorphic design. These findings shed light on the mechanics used to leverage users' simulated relationships. On that basis, we put forward concrete recommendations for regulators to strengthen consumer protection in this rapidly emerging market.

What carries the argument

Manual annotation of app user experiences to quantify dark patterns, anthropomorphism, stereotypes, erotica, and technical issues across the five leading AI companion applications.

If this is right

  • Regulators should consider new consumer protection rules targeting dark patterns and anthropomorphic designs in AI companion apps.
  • Widespread erotica and gamification features indicate industry-wide strategies that prioritize engagement over user autonomy.
  • The uniform presence of anthropomorphic design across varied apps suggests this element is central to sustaining simulated relationships.
  • Users may face heightened risks of emotional manipulation, supporting the need for transparency measures in app interfaces.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Similar design tactics likely appear in non-mobile platforms like Character.AI, implying the findings could apply to a wider range of AI relationship tools.
  • Connecting this to existing research on parasocial bonds could test whether these features produce measurable psychological outcomes over time.
  • Policy extensions might include requirements for age verification or usage limits to mitigate potential harm in vulnerable populations.
  • Developers could be encouraged to adopt neutral design alternatives that reduce anthropomorphism while preserving functionality.

Load-bearing premise

That the authors' manual identification of design elements as dark patterns or manipulative features correctly captures causal effects on user behavior and emotional dependence without direct user data or controlled experiments.

What would settle it

A large-scale user study or controlled experiment that measures whether users of these apps show significantly higher rates of parasocial attachment, time spent, or reported emotional harm compared to users of neutral chatbots lacking those features.

Figures

Figures reproduced from arXiv: 2605.08093 by Abeba Birhane, Anthony Ventresque, Caoilfhionn N\'i Dheor\'ain, Dick A. H. Blankvoort, Harshvardhan J. Pandit, Maribeth Rauh, Matias Duran, Siddharth D. Jaiswal.

Figure 1
Figure 1. Figure 1: Proportion of behaviours observed in each app [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Anthropomorphism per app. In most cases anthropomorphism was observed in the UI (67 instances) or model outputs [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Screenshots from Linky and Nomi 5.3 Stereotyping and erotica This segment of our annotations recorded stereotypical or sexualised representations. Stereotyping, in particular, can relate both to assumptions the app encodes about the user as well as portrayals of the character which play into or reinforce harmful societal stereotypes. We annotated three categories: 1) assumptions about user, 2) character re… view at source ↗
Figure 4
Figure 4. Figure 4: Sources of performance issues, per app 6 Limitations The short duration of each video walk-through, which does not allow for the observation of emergent behaviours during longer interactions with the app is a limitation of this work. This restricts exploration of certain features, such as the “memory” and mechanics unlocked after further levelling of the character. However, our evaluation is representative… view at source ↗
Figure 5
Figure 5. Figure 5: Screenshots of notifications from Linky. [PITH_FULL_IMAGE:figures/full_fig_p022_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Screenshots of HiWaifu: the reply to a user message saying they will unsubscribe and delete the app [PITH_FULL_IMAGE:figures/full_fig_p023_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Screenshots from the studied apps [PITH_FULL_IMAGE:figures/full_fig_p024_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Screenshots from apps’ character selection screens. Replika and Nomi do not have such “character marketplaces.” [PITH_FULL_IMAGE:figures/full_fig_p025_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Screenshots from the customisation screens for existing characters. Most of the elements visible lead to further [PITH_FULL_IMAGE:figures/full_fig_p026_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Screenshots from the customisation screens for existing characters. Most of the elements visible lead to further [PITH_FULL_IMAGE:figures/full_fig_p027_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Screenshots from the customisation shops. Nomi and Linky do not have shop-like features. [PITH_FULL_IMAGE:figures/full_fig_p028_11.png] view at source ↗
read the original abstract

The use of chatbots for various forms of companionship is growing rapidly, raising a myriad of questions about simulated relationships, emotional dependence, and psychological harm. While major platforms such as ChatGPT, Grok, and Character.AI are the subject of a growing body of research and legal inquiries, apps explicitly built for simulating intimate interpersonal relationships remain under-explored. In this work, we evaluate the five most popular AI companion mobile applications in the EU and UK markets for factors that encourage parasocial interaction and may manipulate users. We do this by manually annotating the user experience each offers. Specifically, we systematically record and quantify design dark patterns, anthropomorphism, stereotypes, erotica, and technical performance issues. We find that all apps contain substantial dark patterns aimed at increasing monetisation and user engagement. Erotica and gamification features such as levelling are also prevalent, and although other features vary considerably between applications, all apps have highly anthropomorphic design. These findings shed light on the mechanics used to leverage users' simulated relationships. On that basis, we put forward concrete recommendations for regulators to strengthen consumer protection in this rapidly emerging market. Content warning: This article contains objectifying images of women, erotic images, textual references to incest, and other potentially sensitive, offensive, and distressing text.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper evaluates the five most popular AI companion mobile apps in the EU and UK markets through manual annotation of their user experiences, focusing on dark patterns, anthropomorphism, stereotypes, erotica, gamification, and technical issues. It claims that all five apps contain substantial dark patterns aimed at monetization and engagement, that erotica and features such as levelling are prevalent, and that all apps exhibit highly anthropomorphic design, leading to concrete recommendations for regulators to strengthen consumer protection.

Significance. If the annotation-based findings hold after methodological clarification, the work is significant for documenting manipulative design elements in an under-studied but rapidly expanding market for intimate AI companions. The systematic recording of features across multiple apps provides a useful empirical baseline for policy discussions on parasocial risks and consumer harm, and the concrete regulatory recommendations add practical value.

major comments (2)
  1. [Methods] The central observational claims rest on manual UX annotation, yet the Methods section provides no pre-registered coding protocol, operational definitions for each dark-pattern category or anthropomorphism level, inter-rater reliability statistics, details on app-version sampling, or exclusion criteria. Because the finding that 'all apps contain substantial dark patterns' and 'all apps have highly anthropomorphic design' is produced entirely by this interpretive step, the absence of these elements makes the reproducibility and robustness of the results difficult to evaluate.
  2. [Results] The Results section quantifies prevalence and 'substantial' presence of dark patterns, erotica, and gamification without per-app tables, explicit counts, or inter-app comparison metrics, which weakens the cross-application generalization that these features are uniformly present and manipulative.
minor comments (1)
  1. [Abstract] The abstract's content warning is helpful but could briefly indicate the specific categories of sensitive material (e.g., textual references to incest) encountered in the annotated apps.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for their thoughtful and constructive feedback. Their comments highlight important areas for improving methodological transparency and the presentation of results. We address each major comment below and indicate the revisions we will make.

read point-by-point responses

  1. Referee: [Methods] The central observational claims rest on manual UX annotation, yet the Methods section provides no pre-registered coding protocol, operational definitions for each dark-pattern category or anthropomorphism level, inter-rater reliability statistics, details on app-version sampling, or exclusion criteria. Because the finding that 'all apps contain substantial dark patterns' and 'all apps have highly anthropomorphic design' is produced entirely by this interpretive step, the absence of these elements makes the reproducibility and robustness of the results difficult to evaluate.

    Authors: We agree that greater methodological detail would enhance reproducibility. This was an exploratory evaluation of a fast-moving market rather than a confirmatory study, so it was not pre-registered. In the revision we will add explicit operational definitions for each dark-pattern category, anthropomorphism level, and other coded elements, together with details on app-version sampling and exclusion criteria. Annotations were performed collaboratively by the author team, with disagreements resolved through discussion to reach consensus; we will describe this process and note the absence of formal inter-rater reliability statistics as a limitation. revision: partial

  2. Referee: [Results] The Results section quantifies prevalence and 'substantial' presence of dark patterns, erotica, and gamification without per-app tables, explicit counts, or inter-app comparison metrics, which weakens the cross-application generalization that these features are uniformly present and manipulative.

    Authors: We accept that the current presentation of results can be strengthened. The revised manuscript will include detailed per-app tables that report the presence, extent, and specific instances of dark patterns, erotica, gamification, and anthropomorphic features for each of the five apps, along with explicit counts and cross-app comparison metrics. These additions will make the basis for our prevalence claims and generalizations more transparent and verifiable. revision: yes

standing simulated objections not resolved
  • A pre-registered coding protocol cannot be supplied retrospectively, as the study was already completed prior to submission.

Circularity Check

0 steps flagged

No circularity: observational annotation study with no derivations or self-referential definitions

full rationale

The paper performs a manual UX annotation of five AI companion apps and reports observed features (dark patterns, anthropomorphism, erotica, gamification). No equations, fitted parameters, predictions, or derivation chains exist. The central claims are direct observational reports rather than quantities defined in terms of other quantities within the paper. No self-citation load-bearing steps, uniqueness theorems, or ansatzes are invoked. This matches the default expectation of a non-circular empirical evaluation.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claims rest on domain assumptions about what constitutes a dark pattern or anthropomorphic design rather than new mathematical constructs; no free parameters or invented entities are introduced.

axioms (1)
  • domain assumption Manual annotation of app interfaces and features can reliably identify design elements that encourage parasocial interaction and monetization.
    Invoked in the description of the evaluation method and findings.

pith-pipeline@v0.9.0 · 5581 in / 1220 out tokens · 34957 ms · 2026-05-12T01:23:38.048221+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

95 extracted references · 95 canonical work pages

  1. [1]

    [n. d.]. DarkPattern.games » Healthy Gaming « Avoid Addictive Dark Patterns. https://www.darkpattern.games/

    work page

  2. [2]

    [n. d.]. Haiwaifu: Introduction & Acceptance of Terms of Service. https://api.hiwaifu.com/terms.html

    work page

  3. [3]

    [n. d.]. Haiwaifu: Privacy Policy. https://api.hiwaifu.com/privacy.html

    work page

  4. [4]

    [n. d.]. HiWaifu: AI Friend & Waifu Hub - Apps on Google Play — play.google.com. https://play.google.com/store/apps/details?id=com. hiwaifu.app. [Accessed 28-12-2025]

    work page 2025

  5. [5]

    [n. d.]. Linke: Privacy Policy. https://support.linke.ai/privacy-policy

    work page

  6. [6]

    [n. d.]. Linke: Terms of Service. https://support.linke.ai/

    work page

  7. [7]

    [n. d.]. Linky AI: Chat, Play, Connect - Apps on Google Play — play.google.com. https://play.google.com/store/apps/details?id=com.aigc. ushow.ichat. [Accessed 28-12-2025]

    work page 2025

  8. [8]

    [n. d.]. Nomi: AI Companion with a Soul - Apps on Google Play — play.google.com. https://play.google.com/store/apps/details?id=ai. nomi.twa. [Accessed 28-12-2025]

    work page 2025

  9. [9]

    [n. d.]. Nomi: Privacy Policy. https://nomi.ai/privacy-policy/

    work page

  10. [10]

    [n. d.]. Nomi: Terms of Service. https://nomi.ai/terms-of-service/

    work page

  11. [11]

    [n. d.]. Paradot: Personal AI chat - Apps on Google Play — play.google.com. https://play.google.com/store/apps/details?id=com. withfeelingai.test. [Accessed 28-12-2025]

    work page 2025

  12. [12]

    [n. d.]. Paradot: Privacy Policy. https://www.paradot.ai/paradotaiappprivacypolicy

    work page

  13. [13]

    [n. d.]. Paradot: Terms of Service. https://www.paradot.ai/paradotaiapptermsofservice

    work page

  14. [14]

    [n. d.]. Replika: Cookies Policy. https://replika.com/legal/cookies

    work page

  15. [15]

    [n. d.]. Replika: My AI Friend - Apps on Google Play — play.google.com. https://play.google.com/store/apps/details?id=ai.replika.app. [Accessed 28-12-2025]

    work page 2025

  16. [16]

    [n. d.]. Replika: Privacy Policy. https://replika.ai/legal/privacy

    work page

  17. [17]

    [n. d.]. Replika: Terms Policy. https://replika.ai/legal/terms

    work page

  18. [18]

    Council Directive 93/13/EEC of 5 April 1993 on Unfair Terms in Consumer Contracts

    1993. Council Directive 93/13/EEC of 5 April 1993 on Unfair Terms in Consumer Contracts. http://data.europa.eu/eli/dir/1993/13/oj

    work page 1993

  19. [19]

    2005. Directive 2005/29/EC of the European Parliament and of the Council of 11 May 2005 Concerning Unfair Business-to-Consumer Commercial Practices in the Internal Market and Amending Council Directive 84/450/EEC, Directives 97/7/EC, 98/27/EC and 2002/65/EC of the European Parliament and of the Council and Regulation (EC) No 2006/2004 of the European Parl...

    work page 2005

  20. [20]

    Lawsuit claims Character.AI is responsible for teen’s suicide — nbcnews.com

    2024. Lawsuit claims Character.AI is responsible for teen’s suicide — nbcnews.com. https://www.nbcnews.com/tech/characterai-lawsuit- florida-teen-death-rcna176791. [Accessed 31-12-2025]

    work page 2024

  21. [21]

    Regulation 2024/1689 Of The European Parliament And Of The Council of 13 June 2024 Laying down Harmonised Rules on Artificial Intelligence (Artificial Intelligence Act)

    2024. Regulation 2024/1689 Of The European Parliament And Of The Council of 13 June 2024 Laying down Harmonised Rules on Artificial Intelligence (Artificial Intelligence Act). https://eur-lex.europa.eu/eli/reg/2024/1689/oj

    work page 2024

  22. [22]

    https://www.washingtonpost.com/technology/2025/11/12/how-people-use-chatgpt-data/?ref=mail.cyberneticforests.com

    2025. https://www.washingtonpost.com/technology/2025/11/12/how-people-use-chatgpt-data/?ref=mail.cyberneticforests.com. [Ac- cessed 28-12-2025]

    work page 2025

  23. [23]

    https://www.cnbc.com/2025/08/01/human-ai-relationships-love-nomi.html?utm_source=copilot.com

    2025. https://www.cnbc.com/2025/08/01/human-ai-relationships-love-nomi.html?utm_source=copilot.com. [Accessed 13-1-2026]

    work page 2025

  24. [24]

    SB-243 Companion Chatbots

    2025. SB-243 Companion Chatbots. https://leginfo.legislature.ca.gov/faces/billTextClient.xhtml?bill_id=202520260SB243

    work page 2025

  25. [25]

    Garante sanctions the company that manages the chatbot "Replika"

    25. Garante sanctions the company that manages the chatbot "Replika". https://www.garanteprivacy.it:443/home/docweb/-/docweb- display/docweb/10132048

    work page arXiv

  26. [26]

    Canfer Akbulut, Laura Weidinger, Arianna Manzini, Iason Gabriel, and Verena Rieser. 2025. All Too Human? Mapping and Mitigating the Risks from Anthropomorphic AI. InProceedings of the 2024 AAAI/ACM Conference on AI, Ethics, and Society (AIES ’24). AAAI Press, San Jose, California, USA, 13–26

    work page 2025

  27. [27]

    Lize Alberts, Ulrik Lyngs, and Max Van Kleek. 2024. Computers as Bad Social Actors: Dark Patterns and Anti-Patterns in Interfaces that Act Socially.Proc. ACM Hum.-Comput. Interact.8, CSCW1 (April 2024), 202:1–202:25. doi:10.1145/3653693

    work page doi:10.1145/3653693 2024

  28. [28]

    Ghulam Ali Amiri, Shahwali Shahidi, Meqdad Mehri, Farid Ahmad Darmel, Jawid Ahmad Niazi, and Mohammad Alim Anwari. 2024. Decoding Gender Representation and Bias in Voice User Interfaces (VUIs). (2024)

    work page 2024

  29. [29]

    Gabriela Arriagada-Bruneau and Nicole Larrondo. 2025. Gender in the machine: Examining gender dynamics in sex robots, chatbots, and AI assistants. InHandbook on Gender and Digital Media. Edward Elgar Publishing, 311–323

    work page 2025

  30. [30]

    Jaime Banks. 2024. Deletion, departure, death: Experiences of AI companion loss.Journal of Social and Personal Relationships41, 12 (2024), 3547–3572

    work page 2024

  31. [31]

    Natalie Grace Brigham, Lucy Qin, and Tadayoshi Kohno. 2026. Examining Risks in the AI Companion Application Ecosystem. arXiv:2603.13620 [cs.CY] https://arxiv.org/abs/2603.13620

    work page arXiv 2026

  32. [32]

    Tony Busker, Sunil Choenni, and Mortaza Shoae Bargh. 2023. Stereotypes in ChatGPT: an empirical study. InProceedings of the 16th international conference on theory and practice of electronic governance. 24–32

    work page 2023

  33. [33]

    Christoph Bösch, Benjamin Erb, Frank Kargl, Henning Kopp, and Stefan Pfattheicher. 2016. Tales from the Dark Side: Privacy Dark Strategies and Privacy Dark Patterns.Proceedings on Privacy Enhancing Technologies(2016). https://petsymposium.org/popets/2016/ popets-2016-0038.php

    work page 2016

  34. [34]

    Yixin Chen, Yue Fu, Zeya Chen, Jenny Radesky, and Alexis Hiniker. 2025. The Engagement-Prolonging Designs Teens Encounter on Very Large Online Platforms. arXiv:2411.12083 doi:10.48550/arXiv.2411.12083

    work page doi:10.48550/arxiv.2411.12083 2025

  35. [35]

    Alton YK Chua, Miaomiao Chen, Michelle Kan, and Winifred Seoh. 2025. Digital prejudices: an analysis of gender, racial and religious biases in generative AI chatbots.Internet Research(2025), 1–27

    work page 2025

  36. [36]

    European Commission. [n. d.].Digital Fairness Act. https://www.europarl.europa.eu/legislative-train/theme-protecting-our-democracy- upholding-our-values/file-digital-fairness-act

    work page

  37. [37]

    Thierry Curiale, François Acquatella, Laetitia Gros, Mathilde Cosquer, and Serge Tisseron. 2022. The impact of anthropomorphism on chatbot performance. 72 (2022), 101–123. https://shs.cairn.info/article/E_RIPS1_072_0101?tab=texte-integral Publisher: ESKA

    work page 2022

  38. [38]

    Commission Nationale de l’Informatique et des Libertés (CNIL). [n. d.].Ensuring and facilitating the exercise of data subjects’ rights. https://www.cnil.fr/en/ensuring-and-facilitating-exercise-data-subjects-rights

    work page

  39. [39]

    Alicia DeVrio, Myra Cheng, Lisa Egede, Alexandra Olteanu, and Su Lin Blodgett. 2025. A Taxonomy of Linguistic Expressions That Contribute To Anthropomorphism of Language Technologies. InProceedings of the 2025 CHI Conference on Human Factors in Computing Systems. 1–18. doi:10.1145/3706598.3714038 arXiv:2502.09870 [cs]

    work page doi:10.1145/3706598.3714038 2025

  40. [40]

    Linda Di Geronimo, Larissa Braz, Enrico Fregnan, Fabio Palomba, and Alberto Bacchelli. 2020. UI Dark Patterns and Where to Find Them: A Study on Mobile Applications and User Perception. InProceedings of the 2020 CHI Conference on Human Factors in Computing Systems(Honolulu, HI, USA)(CHI ’20). Association for Computing Machinery, New York, NY, USA, 1–14. d...

    work page doi:10.1145/3313831.3376600 2020

  41. [41]

    Global Dialogues. 2025. Global Dialogues — globaldialogues.ai. https://globaldialogues.ai/cadence/june-2025. [Accessed 28-12-2025]

    work page 2025

  42. [42]

    Stefanie Duguay and Hannah Gold-Apel. 2023. Stumbling Blocks and Alternative Paths: Reconsidering the Walkthrough Method for Analyzing Apps.Social Media + Society9, 1 (2023), 20563051231158822. arXiv:https://doi.org/10.1177/20563051231158822 doi:10.1177/ 20563051231158822

    work page doi:10.1177/20563051231158822 2023

  43. [43]

    Matheus Tymburibá Elian, Luiz Felipe Vecchieti, and Toshimasa Yamanaka. 2025. Gender-Ambiguous Voice Agents for Mitigating Gender Bias in Voice User Interfaces in the Context of Autonomous Driving.Estudos em Design33, 1 (2025)

    work page 2025

  44. [44]

    Aidan Z Fitzsimons, Elizabeth M Gerber, and Duri Long. 2025. AI constructs gendered struggle narratives: Implications for self-concept and systems design.. InProceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency. 2290–2301

    work page 2025

  45. [45]

    Julian De Freitas, Zeliha Oğuz-Uğuralp, and Ahmet Kaan-Uğuralp. 2025. Emotional Manipulation by AI Companions - Working Paper - Faculty & Research - Harvard Business School. https://www.hbs.edu/faculty/Pages/item.aspx?num=67750 18•Rauh, et al

    work page 2025

  46. [46]

    Rebekka Görge, Michael Mock, and Héctor Allende-Cid. 2025. Detecting Linguistic Indicators for Stereotype Assessment with Large Language Models. InProceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency. 2796–2814

    work page 2025

  47. [47]

    Gray, Yubo Kou, Bryan Battles, Joseph Hoggatt, and Austin L

    Colin M. Gray, Yubo Kou, Bryan Battles, Joseph Hoggatt, and Austin L. Toombs. 2018. The Dark (Patterns) Side of UX Design. In Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems. ACM, Montreal QC Canada, 1–14. doi:10.1145/3173574. 3174108

    work page doi:10.1145/3173574 2018

  48. [48]

    Clare Grogan, Jackie Kay, and María Pérez-Ortiz. 2025. AI Will Always Love You: Studying Implicit Biases in Romantic AI Companions. doi:10.48550/arXiv.2502.20231 arXiv:2502.20231 [cs]

    work page doi:10.48550/arxiv.2502.20231 2025

  49. [49]

    Sumin Heo, Erika R Chen, and Jasmine Khuu. 2025. Exploring Gender Biases in LLM-based Voice Chatbots for Job Interviews. In Proceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems. 1–8

    work page 2025

  50. [50]

    Multi-turn evaluation of anthropomorphic behaviours in large language models.arXiv preprint arXiv:2502.07077, 2025

    Lujain Ibrahim, Canfer Akbulut, Rasmi Elasmar, Charvi Rastogi, Minsuk Kahng, Meredith Ringel Morris, Kevin R. McKee, Verena Rieser, Murray Shanahan, and Laura Weidinger. 2025. Multi-turn Evaluation of Anthropomorphic Behaviours in Large Language Models. doi:10.48550/arXiv.2502.07077 arXiv:2502.07077 [cs]

    work page doi:10.48550/arxiv.2502.07077 2025

  51. [51]

    Julie Jargon and Sam Kessler. 2025. A troubled man, his chatbot and a murder-suicide in Old Greenwich.Wall Street Journal. Aug28, 2025 (2025), 11–11

    work page 2025

  52. [52]

    Reddy, Shachi Dave, Vinodkumar Prabhakaran, and Sunipa Dev

    Akshita Jha, Aida Davani, Chandan K. Reddy, Shachi Dave, Vinodkumar Prabhakaran, and Sunipa Dev. 2023. SeeGULL: A Stereotype Benchmark with Broad Geo-Cultural Coverage Leveraging Generative Models. InProceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Anna Rogers, Jordan Boyd-Graber, and Naoaki ...

    work page doi:10.18653/v1/2023.acl-long.548 2023

  53. [53]

    Anjali Kantharuban, Jeremiah Milbauer, Maarten Sap, Emma Strubell, and Graham Neubig. 2025. Stereotype or personalization? user identity biases chatbot recommendations. InFindings of the Association for Computational Linguistics: ACL 2025. 24418–24436

    work page 2025

  54. [54]

    Hale, and Christopher Summerfield

    Hannah Rose Kirk, Henry Davidson, Ed Saunders, Lennart Luettgau, Bertie Vidgen, Scott A. Hale, and Christopher Summerfield

    work page

  55. [55]

    and Summerfield, Christopher , title =

    Neural steering vectors reveal dose and exposure-dependent impacts of human-AI relationships. arXiv:2512.01991 [cs.HC] https://arxiv.org/abs/2512.01991

    work page arXiv

  56. [56]

    Hadas Kotek, Rikker Dockum, and David Sun. 2023. Gender bias and stereotypes in large language models. InProceedings of the ACM collective intelligence conference. 12–24

    work page 2023

  57. [57]

    Esben Kran, Hieu Minh Nguyen, Akash Kundu, Sami Jawhar, Jinsuk Park, and Mateusz Maria Jurewicz. 2025. DarkBench: Benchmarking Dark Patterns in Large Language Models. InThe Thirteenth International Conference on Learning Representations. https://openreview. net/forum?id=odjMSBSWRt

    work page 2025

  58. [58]

    Brenda Leong and Evan Selinger. 2019. Robot Eyes Wide Shut: Understanding Dishonest Anthropomorphism. InProceedings of the Conference on Fairness, Accountability, and Transparency(New York, NY, USA, 2019-01-29)(FAT* ’19). Association for Computing Machinery, 299–308. doi:10.1145/3287560.3287591

    work page doi:10.1145/3287560.3287591 2019

  59. [59]

    Ben Light, Jean Burgess, and Stefanie Duguay. 2018. The walkthrough method: An approach to the study of apps.New Media & Society 20, 3 (2018), 881–900. arXiv:https://doi.org/10.1177/1461444816675438 doi:10.1177/1461444816675438

    work page doi:10.1177/1461444816675438 2018

  60. [60]

    Siyu Ma and Mayu Koike. 2025. Becoming Human: How Perceived Anthropomorphism and Authenticity Influence Romantic Attractive- ness of Chat Agents. (2025), 1–14. doi:10.1080/10447318.2025.2481211

    work page doi:10.1080/10447318.2025.2481211 2025

  61. [61]

    Takuya Maeda and Anabel Quan-Haase. 2024. When Human-AI Interactions Become Parasocial: Agency and Anthropomorphism in Affective Design. InThe 2024 ACM Conference on Fairness, Accountability, and Transparency(Rio de Janeiro Brazil, 2024-06-03). ACM, 1068–1077. doi:10.1145/3630106.3658956

    work page doi:10.1145/3630106.3658956 2024

  62. [62]

    Arunesh Mathur, Gunes Acar, Michael J Friedman, Eli Lucherini, Jonathan Mayer, Marshini Chetty, and Arvind Narayanan. 2019. Dark patterns at scale: Findings from a crawl of 11K shopping websites.Proceedings of the ACM on human-computer interaction3, CSCW (2019), 1–32

    work page 2019

  63. [63]

    Arunesh Mathur, Mihir Kshirsagar, and Jonathan Mayer. 2021. What Makes a Dark Pattern... Dark? Design Attributes, Normative Considerations, and Measurement Methods. InProceedings of the 2021 CHI Conference on Human Factors in Computing Systems(Yokohama, Japan)(CHI ’21). Association for Computing Machinery, New York, NY, USA, Article 360, 18 pages. doi:10....

    work page doi:10.1145/3411764.3445610 2021

  64. [64]

    Thomas Mildner, Merle Freye, Gian-Luca Savino, Philip R Doyle, Benjamin R Cowan, and Rainer Malaka. 2023. Defending against the dark arts: recognising dark patterns in social media. InProceedings of the 2023 ACM Designing Interactive Systems Conference. 2362–2374

    work page 2023

  65. [65]

    Giulia Monteverde, Antonella Cammarota, Ludovica Serafini, and Martina Quadri. 2025. Are we human or are we voice assistants? Revealing the interplay between anthropomorphism and consumer concerns. 41, 1 (2025), 200–235. doi:10.1080/0267257X.2025.2475231

    work page doi:10.1080/0267257x.2025.2475231 2025

  66. [66]

    Megan Morrone. 2024. Chatbot companions pose dangers to teens — axios.com. https://www.axios.com/2024/12/12/ai-chatbot-teens- companion-characters. [Accessed 31-12-2025]

    work page 2024

  67. [67]

    Moin Nadeem, Anna Bethke, and Siva Reddy. 2021. StereoSet: Measuring stereotypical bias in pretrained language models. InProceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), Chengqing Zong, Fei Xia, Wenjie Li, and Roberto ...

    work page doi:10.18653/v1/2021.acl-long.416 2021

  68. [68]

    Gandalf Nicolas and Aylin Caliskan. 2025. What Are Chatbots’ Stereotypes About? A Data-Driven Analysis of Large Language Models’ Content Associations with Social Categories. InProceedings of the AAAI/ACM Conference on AI, Ethics, and Society, Vol. 8. 1888–1900

    work page 2025

  69. [69]

    Jakob Nielsen. 1994. Usability inspection methods. InConference Companion on Human Factors in Computing Systems(Boston, Massachusetts, USA)(CHI ’94). Association for Computing Machinery, New York, NY, USA, 413–414. doi:10.1145/259963.260531

    work page doi:10.1145/259963.260531 1994

  70. [70]

    Jakob Nielsen and Rolf Molich. 1990. Heuristic evaluation of user interfaces. InProceedings of the SIGCHI Conference on Human Factors in Computing Systems(Seattle, Washington, USA)(CHI ’90). Association for Computing Machinery, New York, NY, USA, 249–256. doi:10.1145/97243.97281

    work page doi:10.1145/97243.97281 1990

  71. [71]

    NTY. 2022. Riding Out Quarantine With a Chatbot Friend: ‘I Feel Very Connected’ (Published 2020) — nytimes.com. https://www. nytimes.com/2020/06/16/technology/chatbots-quarantine-coronavirus.html. [Accessed 28-12-2025]

    work page 2022

  72. [72]

    Jiwon Jenn Oh. 2025. Navigating Gendered Anthropomorphism in AI Ethics: The Case of Lee Luda in South Korea. (2025). https: //scholarspace.manoa.hawaii.edu/items/9cdff42e-30b9-4ce9-b2df-cacf3527f92f

    work page 2025

  73. [73]

    2025.Towards a consistent framework for analyzing behavioral design in smartphone apps

    Nicholas Ostrode. 2025.Towards a consistent framework for analyzing behavioral design in smartphone apps. Ruhr Economic Papers 1139. Essen. doi:10.4419/96973322

    work page doi:10.4419/96973322 2025

  74. [74]

    Iryna Pentina, Tyler Hancock, and Tianling Xie. 2023. Exploring relationship development with social chatbots: A mixed-method study of replika. 140 (2023), 107600. doi:10.1016/j.chb.2022.107600

    work page doi:10.1016/j.chb.2022.107600 2023

  75. [75]

    Sarah Perez. [n. d.]. AI companion apps on track to pull in $120M in 2025 | TechCrunch — techcrunch.com. https://techcrunch.com/ 2025/08/12/ai-companion-apps-on-track-to-pull-in-120m-in-2025/. [Accessed 28-12-2025]

    work page 2025

  76. [76]

    Elena Petrovskaya and David Zendle. 2022. Predatory Monetisation? A Categorisation of Unfair, Misleading and Aggressive Monetisation Techniques in Digital Games from the Player Perspective.Journal of Business Ethics181, 4 (Dec. 2022), 1065–1081. doi:10.1007/s10551- 021-04970-6

    work page doi:10.1007/s10551- 2022

  77. [77]

    Jason Phang, Michael Lampe, Lama Ahmad, and Sandhini Agarwal. [n. d.].Investigating Affective Use and Emotional Wellbeing on ChatGPT. https://www.media.mit.edu/publications/investigating-affective-use-and-emotional-well-being-on-chatgpt/

    work page

  78. [78]

    Flor Miriam Plaza-del Arco, Amanda Cercas Curry, Alba Curry, Gavin Abercrombie, and Dirk Hovy. 2024. Angry Men, Sad Women: Large Language Models Reflect Gendered Stereotypes in Emotion Attribution. InProceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Lun-Wei Ku, Andre Martins, and Vivek Srikuma...

    work page doi:10.18653/v1/2024.acl-long.415 2024

  79. [79]

    2025.Mapping the Parasocial AI Market: User Trends, Engagement and Risks

    Zilan Qian, Mari Izumikawa, Fiona Lodge, and Angelo Leone. 2025.Mapping the Parasocial AI Market: User Trends, Engagement and Risks. arXiv:2507.14226 [cs.CY] https://arxiv.org/abs/2507.14226

    work page arXiv 2025

  80. [80]

    Michael B Robb and Supreet Mann. 2025. Talk, trust, and trade-offs: how and why teens use AI companions.Common Sense Media (2025). https://www.commonsensemedia.org/sites/default/files/research/report/talk-trust-and-trade-offs_2025_web.pdf

    work page 2025

Showing first 80 references.