Cited by 107Social Robot Interaction and HRIEthics and Social Impacts of AIInnovative Human-Technology Interaction

The impacts of companion AI on human relationships: risks, benefits, and design considerations

Open access
Malfacini, Kim·2025·AI & SOCIETY

Abstract

AI applications intended for companionship have increased in both availability and capabilities in recent years, alongside a spike in loneliness (Johnson 2023 ; Murthy 2023 ). According to a 2024 analysis by venture capital firm Andreessen Horowitz, companion AI makes up 16 of the top 100 AI apps based on web traffic and monthly active users (Moore 2024 ). Moreover, today’s AI applications have a greater command of human language, an increasing capacity for storing memories, and more varied modalities to support an increasing diversity of experiences, a known driver in the development and deepening of emotional relationships (Skjuve et al. 2022 ).

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1007/s00146-025-02318-6

Cites

64 of 64 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • A substantial body of work examines attachment dysfunction, social withdrawal, unrealistic relational expectations, and potential manipulation in human-AI relationships.

    “A substantial body of work examines attachment dysfunction, social withdrawal, unrealistic relational expectations, and the potential for manipulation in human-AI relationships (Turkle, 2011; Malfacini, 2025; Zhou & Zhang, 2025).”

    From Intimate Strangers by Design: A Uses and Gratifications Analysis of AI Companionship (2026) · cites this paper as Malfacini, 2025

  • A substantial body of scholarship has identified the asymmetry of AI companionship as a mechanism through which AI companions may foster unrealistic relational expectations, displace effortful human intimacy, and normalize one-sided emotional dynamics.

    “A substantial body of scholarship has identified this asymmetry as a mechanism through which AI companions may foster unrealistic relational expectations, displace effortful human intimacy, and normalize one-sided emotional dynamics (Turkle, 2011; Malfacini, 2025; Yankouskaya et al., 2025).”

    From Intimate Strangers by Design: A Uses and Gratifications Analysis of AI Companionship (2026) · cites this paper as Malfacini, 2025

  • Since AI companions are endlessly patient, attentive, and always available, this may lead users to expect unrealistically the same in humans.

    “since AI companions are endlessly patient, attentive, and always available, this may lead users to expect (unrealistically) the same in humans (Malfacini, 2025).”

    From Are Attributions of Consciousness to AI Chatbots Epistemically Innocent? (2026) · cites this paper as Malfacini, 2025

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Cites

Flagged first, then by citations

Research works printed but not linked: 6

  1. Braverman H (1998) Labor and monopoly capital: the degradation of work in the Twentieth Century. NYU Press. https:// www. jstor. org/ stable/ j. ctt9q frkf
  2. Finnis, J (1980) Natural Law and Natural Rights. First. Oxford Uni- versity Press. https:// www. amazon. com/ Natur al-Law-Rights-Clare ndon/ dp/ 01987 61104
  3. Gallegos D (2024) 'Is it ok to be mean to a chatbot?' The Wall Street Journal, 15 Feb 2024. https:// www. wsj. com/ tech/ ai/ artifi cial-intel ligen ce-chatb ot-manne rs-65a4e df9
  4. Gardner W L, Pickett CL, Knowles M (2005) 'Social snacking and shielding: using social symbols, selves, and surrogates in the ser- vice of belonging needs'. In The Social Outcast: Ostracism, Social Exclusion, Rejection, and Bullying, 227-41. Sydney Symposium of Social Psychology Series. New York, NY, US: Psychology Press.
  5. Mehan JE (2022) 'Chapter 5: AI and Our Society: Will the Result Be Professional, Ethical, and Social Deskilling, or Upskilling?' In Artificial Intelligence -Ethical, Social, and Security Impacts for the Present and the Future. IT Governance Publishing. https:// www. oreil ly. com/ libra ry/ view/ artifi cial-intel ligen ce/ 97817 87783 720/
  6. Ruane E, Birhane A, Ventresque A (2019) Conversational AI: Social and Ethical Considerations. AICS 2019: 104-115. https:// ceur-ws. org/ Vol-2563/ aics_ 12. pdf

Cited by

Sorted by citations

Authors
Malfacini, K.
Affiliations
OpenAI (United States) (US)
Corresponding author
Malfacini, Kim
Venue
AI & SOCIETY
Published
2025-04-16
Volume
40
Issue
7
Pages
5527-5540
ISSN
  • 0951-5666 (ISSN-L)
  • 1435-5655
OpenAlex
W4409502783
Cited by
105
References
63
Citations by year
2026: 93 · 2025: 12
FWCI
116.34
Type
journal article
Language
en
Open access
Locations
AI & Society (publishedVersion, cc-by)
Topics
  • Social Sciences › Psychology › Social Psychology › Social Robot Interaction and HRI
  • Social Sciences › Safety Research › Ethics and Social Impacts of AI
  • Physical Sciences › Computer Science › Human-Computer Interaction › Innovative Human-Technology Interaction
Keywords
AI companions, social robots, human-robot relationships, loneliness, human-AI interaction

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review