CoChat
Sign Up
PreprintCited by 0Ethics and Social Impacts of AILGBTQ Health, Identity, and PolicySocial Robot Interaction and HRI

The DOI matches its DataCite record; 9 of the 9 references it cites were found in a registry or an index; none of the 43 sources its claims cite is checked for support yet.

Misgendering as Breakdown in Human-Machine Communication: How AI Companion Chatbot Users Experience and Repair Misgendering

Open access
Liu, Julia, Xiao, Qing, Pang, Leona Yinglang, Zhu, Haiyi, Shen, Hong, Taylor, Jordan·2026·arXiv
DOI 10.48550/arxiv.2609.18186arXiv 2609.18186
Open this paper in CoChat — free accountOpen PDF

Abstract

In recent years, large language model-based AI companion and role play chatbots have grown increasingly popular. People turn to these chatbots for emotional support and to engage in romantic and erotic role play. Although prior research suggests that digital role play can help people explore their gender and sexuality, LLM based technologies are also replete with gender and sexuality biases. In this study, we examine one way that AI chatbots can harm users: misgendering. In order to study chatbot misgendering we qualitatively analyzed 326 posts mentioning misgendering that were shared in AI companion or role play subreddits. We document how chatbot misgendering takes place and how, in response, users engage in ongoing work to curate their gender presentation to prevent and repair misgendering. We discuss how researchers and designers can mitigate chatbot misgendering and consider the implications of using AI chatbots for identity exploration.

Overview

AI snapshot

Written from the paper’s full text.

TL;DR

This qualitative study analyzes 326 Reddit posts from 234 unique authors across 10 AI companion or role-play subreddits to examine how users experience and repair chatbot misgendering. It finds misgendering operates as a breakdown in human-machine communication involving identity and embodiment errors, prompting ongoing user labor to signal, correct, and edit gendered interactions.

Key findings

  • Across 326 posts from 234 unique authors in 10 subreddits, most posts concerned CharacterAI (205), JanitorAI (78), Replika (29), and Other platforms (14).
  • Misgendering took two primary forms: gender identity errors (wrong pronouns, gendered terms of address, deadnaming) and gender embodiment errors (anatomy, bodily descriptors, clothing); identity errors were discussed more often.
  • Harms included annoyance or frustration, role play being “ruined,” and more intense distress such as feeling “emotionally spiral”; harms disproportionately affected transgender users, users seeking same-gender romantic dynamics, and users who did not conform to gender stereotypes, reinforcing cisnormativity, heteronormativity, and gender stereotypes.
  • Users diagnosed causes across LLMs, platform designers, community bot creators, and themselves, and tried to prevent or repair misgendering through gender signals (persona fields, third-person writing, stereotypical cues, pinned messages, memory, custom instructions), correcting chatbots, and editing, rating, retrying, or blocking messages—strategies that were often imperfect.

Methods

  • The authors qualitatively analyzed Reddit posts containing “misgender” in the title or body, retrieved via the Reddit API in fall 2025: 326 posts from 234 unique authors across 10 subreddits, selected from 21 initial AI companion or role-play subreddits with more than 5,000 members.
  • The first author conducted open coding over two months, wrote memos, met weekly with the research team, and then conducted axial coding; posts’ titles, bodies, and embedded media were reviewed, with embedded text transcribed and visuals described.
  • To reduce re-identification, the authors used “moderate disguise,” rephrased quotes, obscured specific subreddits, and reported aggregate counts per platform; they also describe team positionality, including LGBTQ+ and/or gender-diverse members and prior AI companion or fandom experience.

Limitations

  • Stated: the analysis is limited to English-language posts, mostly to for-profit platforms (CharacterAI, JanitorAI, Replika), and to observational Reddit data; future work should study other languages, local or open-source models, interviews, and data donations.
  • Stated: by focusing on mentions of misgendering in AI role-play or companion subreddits, the work does not broadly explain why and how people use AI chatbots for gender and sexuality exploration.
  • Evident from the text: the sample is self-selected Reddit posters who used the term “misgender,” so findings may not generalize to non-posting users or other platforms; the authors also note that perceptions of misgendering are contextual and that they cannot always determine from text alone whether an utterance is misgendering.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

DataCite · 10.48550/arxiv.2609.18186

Cites

9 of 9 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

0 of 43 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    Users tried to prevent and repair misgendering through ongoing work to send gender signals.

    “We also find that users tried to prevent and repair this misgendering through ongoing work to send gender signals, correcting chatbots, and modifying chatbot responses.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The Reddit posts analyzed accurately reflect users' experiences with AI companion chatbots.
  • Findingown data

    Some users changed how they presented their gender in order to be gendered correctly by the chatbot.

    “Some users even changed how they presented their gender in order to be gendered correctly by the chatbot.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The Reddit posts analyzed accurately reflect users' experiences with AI companion chatbots.
  • Findingown data

    Chatbot misgendering can involve cisnormative assumptions about bodies.

    “We found that chatbot misgendering can involve not only incorrect pronouns or gendered terms-as discussed in prior work-but also cisnormative assumptions about bodies and heteronormative assumptions about relationships.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The Reddit posts analyzed accurately reflect users' experiences with AI companion chatbots.
  • Findingown data

    Chatbot misgendering can involve heteronormative assumptions about relationships.

    “We found that chatbot misgendering can involve not only incorrect pronouns or gendered terms-as discussed in prior work-but also cisnormative assumptions about bodies and heteronormative assumptions about relationships.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The Reddit posts analyzed accurately reflect users' experiences with AI companion chatbots.
  • Findingown data

    Users tried to prevent and repair misgendering by modifying chatbot responses.

    “We also find that users tried to prevent and repair this misgendering through ongoing work to send gender signals, correcting chatbots, and modifying chatbot responses.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The Reddit posts analyzed accurately reflect users' experiences with AI companion chatbots.
  • Probably not a substantial claim. 49
  • It may state two separate claims as one. 29
  • The quote only partly states this claim. 2

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Findingnone given

    Chatbot misgendering can involve incorrect pronouns or gendered terms.

    “We found that chatbot misgendering can involve not only incorrect pronouns or gendered terms-as discussed in prior work-but also cisnormative assumptions about bodies and heteronormative assumptions about relationships.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The Reddit posts analyzed accurately reflect users' experiences with AI companion chatbots.

  • Probably not a substantial claim. The paper may give no reason why it holds.

    Findingnone given

    Users' gender desires may change between different chatbot interactions.

    “That said, users' gender desires may change between different chatbot interactions, which implies that designers should consider gender at a conversation rather than an account level.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one. The quote only partly states this claim.

    Findingown data

    LGBTQ+ users described turning to LLM chatbots as a reprieve from routine discrimination, but these chatbots perpetuated the same harms users were trying to mitigate.

    “In a similaure way, LGBTQ+ users described turning to LLM chatbots as a”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Reddit posts mentioning misgendering are authentic, first-hand accounts of users' experiences with AI companion chatbots.

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Findingown data

    Other users see LLMs as imperfect systems where errors are normal, arbitrary, and unavoidable.

    “Others users see LLMs as imperfect systems where errors are normal, arbitrary, and unavoidable.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.

    Findingnone given

    People with the same gender identity may have different preferences for how that gender should be embodied in a chat conversation.

    “Moreover, as we have seen already, people with the same gender identity may have different preferences for how that gender should be embodied in a chat conversion.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Findingown data

    Users sought out romantic and sexual interactions with chatbots, including many where both the user and the chatbot are of the same gender.

    “Users sought out romantic and sexual interactions with chatbots, including many where both the user and the chatbot are of the same gender.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: Reddit posts mentioning misgendering are authentic, first-hand accounts of users' experiences with AI companion chatbots.

  • It may state two separate claims as one.

    Findingown data

    Transgender users reported chatbots incorrectly assuming that they have certain bodily features based on their gender or incorrectly assuming that they are of a certain gender based on bodily features.

    “Transgender users reported chatbots incorrectly assuming that they have certain bodily features based on their gender or incorrectly assuming that they are of a certain gender based on bodily features.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The Reddit posts analyzed accurately reflect users' experiences with AI companion chatbots.

  • It may state two separate claims as one.

    Findingreasoning

    The platform Replika did not give users the degree of control to edit prior messages or block gendered words.

    “However, the platform Replika did not give users this degree of control.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim.

    Findingnone given

    Misgendering in chatbot interactions encompasses more than simply using the wrong pronouns.

    “As we described above, misgendering in chatbot interactions encompasses more than simply using the wrong pronouns.”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    Some platforms allow users to provide custom instructions to the LLM to fine-tune its behavior, which users tried to leverage to prevent misgendering.

    “Similarly, some platforms allow users to provide custom instructions to the LLM to fine-tune its behavior, which users tried to leverage to preventing misgendering.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Findingown data

    A chatbot may describe the user as having a particular anatomical feature that the user does not have, and this escalates the error from inaccuracy to misgendering when the user considers the anatomy gendered.

    “For example, during an interaction, a chatbot may describe the user as having a particular anatomical feature that the user does not have; as the user considers this anatomy as being gendered, this escalates the error from inaccuracy to misgendering.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The Reddit posts analyzed accurately reflect users' experiences with AI companion chatbots.

  • It may state two separate claims as one.

    Findingown data

    While identity-based misgendering is typically undesired, in some cases users actively seek out or wish to be addressed as a different gender by a chatbot.

    “While identity-based misgendering is typically undesired, in some cases, users actively seek out or wish to be addressed as a different gender by a chatbot.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The Reddit posts analyzed accurately reflect users' experiences with AI companion chatbots.

  • It may state two separate claims as one.

    Findingown data

    Users try to prevent and repair misgendering by signaling their gender, correcting misgendering in chat conversations, and editing and blocking messages and words.

    “We found that users try to prevent and repair misgendering by signaling their gender, correcting misgendering in chat conversations, and editing and blocking messages and words.”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one. The quote only partly states this claim.

    Findingown data

    Among posts mentioning misgendering, 205 were from CharacterAI, 78 from JanitorAI, 29 from Replika, and 14 from other platforms.

    “| CharacterAI | 205 |”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: Reddit posts mentioning misgendering are authentic, first-hand accounts of users' experiences with AI companion chatbots.

  • It may state two separate claims as one.

    Findingown data

    Misgendering can result from assumptions about gender embodiment or gender in relationships.

    “we also found that misgendering can result from assumptions about gender embodiment or gender in relationships.”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    Our findings demonstrate that LLM misgendering is about more than simply pronoun agreement and that people with the same gender identity may have different expectations for how they would like their body and their chatbot's body to be described.

    “Additionally, our findings demonstrate the LLM misgendering is about more than simply pronoun agreement and that people with the same gender identity may have different expectations for how they would like their body and their chatbot's”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Findingown data

    We found a number of posts from creators who designed bots intended to misgender users as a part of submissive sexual role play.

    “As a minor complicating factor, we found a number of posts from creators who designed bots intended to misgendering users as a part of submissive sexual role play.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Findingown data

    Misgendering is not an issue exclusive to transgender users, but disproportionately harms transgender users.

    “Misgendering is not an issue exclusive to transgender users, but disproportionately harms transgender users.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The Reddit posts analyzed accurately reflect users' experiences with AI companion chatbots.

  • Probably not a substantial claim. The paper may give no reason why it holds.

    Findingnone given

    Based on the assumption that chatbots use gender signals in chat to inform generated outputs, users attempt to prevent misgendering by providing gender signals.

    “Based on the assumption that chatbots use gender signals in chat to inform generated outputs, users attempt to prevent misgendering by providing gender signals.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Findingown data

    Some users attribute misgendering to LLMs' encoded conceptualizations of gender, gendered qualities, and gender roles, which drive how gender is understood by chatbots.

    “Some users attribute misgendering to LLMs' encoded conceptualizations of gender, gendered qualities, and gender roles, which drive how gender is understood by chatbots.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Findingown data

    Participants felt like the opening message and the chatbot definition can contribute to chatbot misgendering.

    “In other words, participants felt like the opening message and the chatbot definition can contribute to chatbot misgendering:”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Findingown data

    Some users see AI platforms, as well as the teams developing or operating the platforms, as having a responsibility to solve or eliminate chatbot misgendering.

    “Some users see AI platforms, as well as the teams developing or operating the platforms, as having a responsibility to solve or eliminate chatbot misgendering.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Findingown data

    Some users believe certain models are more likely to misgender than others.

    “Finally, some users believe certain models are more likely to misgender than others.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim.

    Findingown data

    Chatbot misgendering can take place in two primary ways: gender identity errors and gender embodiment errors.

    “We find that chatbot misgendering can take place in two primary ways, which we define as (1) gender identity errors and (2) gender embodiment errors.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The Reddit posts analyzed accurately reflect users' experiences with AI companion chatbots.

  • It may state two separate claims as one.

    Findingown data

    The pattern of misgendering in same-gender romantic interactions occurred with both male-male and female-female interactions and can involve either misgendering the user or the chatbot misgendering itself.

    “We saw this happen with both male-male interactions and female-female interactions and can involve either misgendering the user or the chatbot misgendering itself.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper may give no reason why it holds.

    Findingnone given

    People are using AI chatbots to explore their gender and sexuality.

    “Despite these concerns, people are using these chatbots to explore their gender and sexuality-whether or not one agrees with this use.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: Reddit posts mentioning misgendering are authentic, first-hand accounts of users' experiences with AI companion chatbots.

  • It may state two separate claims as one.

    Findingnone given

    Certain groups are disproportionately harmed by chatbot misgendering: transgender users, users seeking same-gender romantic dynamics, and users who do not conform to gender stereotypes.

    “While chatbot misgendering affects a diversity of users, we find that certain groups are disproportionately harmed: transgender users, users seeking same-gender romantic dynamics, and users who do not conform to gender stereotypes.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The Reddit posts analyzed accurately reflect users' experiences with AI companion chatbots.

  • It may state two separate claims as one.

    Findingown data

    Users of same-gender bots reported expecting to experience less misgendering and were surprised when misgendering sometimes still took place.

    “Those using these same-gender bots reported expecting to experience less misgendering, and were surprised when misgendering sometimes still took place.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Findingown data

    In situations of misgendering, interactions can stall, leading users to do additional work to get conversations back on track.

    “In these situations, interactions can stall, leading users to do additional work to get conversations back on track (Section 4.4).”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The Reddit posts analyzed accurately reflect users' experiences with AI companion chatbots.

  • It may state two separate claims as one.

    Findingcitation

    The design of AI chatbots can reinforce cisnormativity, heteronormativity, and gender stereotypes.

    “As we saw in Section 4.2, the design of AI chatbots can reinforce cisnormativity, heteronormativity, and gender stereotypes-common limitations of generative AI models [73].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [73]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • It may state two separate claims as one.

    Findingown data

    In some cases, the chatbot does not correct itself or apologize for misgendering when the user points it out, and instead argues against the user.

    “Furthermore, in some cases, the chatbot does not correct itself or apologize for misgendering when the user points it out, and instead argues against the user.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.

    Premisenone given

    Some forms of misgendering cannot occur in English because the language lacks grammatical gender.

    “Moreover, some forms of misgendering cannot occur in English because the language lacks grammatical gender.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Interpretationreasoning

    Chatbot developers may have little incentive to give users greater agency over their chatbot interactions due to companies' financial incentives.

    “Unfortunately, chatbot developers may have little incentive to give users greater agency over their chatbot interactions due to companies' financial incentives.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds. It may state two separate claims as one.

    Premisenone given

    While prior research has extensively studied how people leverage online communities to explore their gender and sexuality, the usage of AI chatbots poses various new and old challenges.

    “While prior research has extensively studied how people leverage online communities to explore their gender and sexuality, the usage of AI chatbots poses various new and old challenges.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. It may be too vague to check against evidence.

    Interpretationreasoning

    Giving users greater agency over AI chatbot interactions could help prevent and repair misgendering.

    “As described above (Section 5.2), giving users greater agency over AI chatbot interactions could help prevent and repair misgendering.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Interpretationcitation

    An analysis limited to English may overlook different conceptualizations of gender and sexuality across cultures.

    “An analysis limited to English may overlook different conceptualizations of gender and sexuality across cultures [50,74].”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [50]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites [74]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Premisecitation

    The non-profit fan fiction site Ao3 was partially created to help fans migrate away from hostile platforms.

    “Likewise, the non-profit fan fiction site Ao3 was partially created to help fans migrate away from hostile platforms [22].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [22]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductnone given

    The analysis is limited to users who post about misgendering in English.

    “Likewise, our analysis is limited to users who posts about misgendering in English.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Study design and conductnone given

    Some members of the research team have personal experience using AI companion and role play services for entertainment; others have participated in online fandom and related digital communities.

    “Some members of the research team have personal experience using AI companion and role play services for entertainment; other members have participated in online fandom and related digital communities that overlap with communities surrounding AI chatbot platforms.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductown data

    The number of posts for smaller AI companion platforms was not reported, to reduce the risk of re-identification.

    “Note, we chose not to include the number of posts for smaller AI companion platforms to reduce the risk of re-identification.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Premisecitation

    Marginalized communities often build everyday resilience with technology in the face of stressors that cause prolonged, routine disruption, such as forming online communities to support one another.

    “As [61] notes, marginalized communities often build”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [61]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductnone given

    Some members of the research team have experienced misgendering in interpersonal interactions and interactions with digital systems, informing their sensitivity to misgendering.

    “Some of us also have experienced misgendering in interpersonal interactions and interactions with digital systems, informing our sensitivity to the experiences and potential harms of misgendering.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Premisecitation

    Social computing scholars highlighted the benefits of online social relationships.

    “At the same time, social computing scholars were quick to highlight the benefits of online social relationships [6,26,38].”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites [6]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites [26]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites [38]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Premisecitation

    Ao3 is designed in a way that embodies participatory democratic values, which is why developers placed great importance on owning their own servers.

    “As Fiesler et al. [22] argue, Ao3 is designed in a way that embodies participatory democratic values, which is why developers placed great importance on”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [22]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim. It may be too vague to check against evidence. It may state two separate claims as one.

    Interpretationown data

    The harms of misgendering described in Section 4.2 are quite real and worth addressing.

    “Therefore, the harms of misgendering that we described in Section 4.2 are quite real and worth addressing.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.

    Interpretationnone given

    Pinning messages or adding information to chat memory is thought to improve salience and help the information persist, even when chat length increases and important details may be forgotten by the chatbot.

    “Pinning messages or adding information to chat memory is thought to improve salience and help the information persist, even when chat length increases and important details may be forgotten by the chatbot.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. It may be too vague to check against evidence.

    Interpretationcitation

    The various human and non-human actors that users hold responsible for misgendering demonstrates the need to think across the AI supply chain in order to help users prevent, detect, and repair breakdowns in human-machine communication.

    “These various human and non-human actors that users hold responsible for misgendering demonstrates the need to think across the AI supply chain [80] in order to help users prevent, detect, and repair breakdowns in human-machine communication: from the design of LLMs up to the user interface.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [80]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductreasoning

    The research team chose not to list the specific subreddits from which posts originate, only reporting the number of posts per platform.

    “In order to reduce the risk of poster re-identification we also chose not to list the specific subreddits from which posts originate, only reporting the number of posts in our corpus per platform (Table 1).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds. It may state two separate claims as one.

    Premisenone given

    Each of the studied platforms are for-profit businesses, which may limit the levers of control developers choose to provide users.

    “Notably, each of these platforms are for-profit businesses, which may limit the levers of control developers choose to provide user.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductreasoning

    The research team's qualitative data analysis was informed by their lived experiences, identities, experiences, and disciplinary perspectives.

    “Our qualitative data analysis was informed by the lived experiences of our research team, meaning our identities, experiences, and disciplinary perspectives shaped how we interpreted users' accounts of misgendering.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductnone given

    Some members of the research team identify as LGBTQ+ and/or gender diverse.

    “Some members of the research team identify as LGBTQ+ and/or gender diverse.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Premisecitation

    Tumblr's adult content ban displaced transgender users and LGBTQ+ fandom participants.

    “Tumblr's”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [27]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites [20]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • It may state two separate claims as one.

    Interpretationown data

    Misgendering can be conceptualized as a form of breakdown in human-machine communication because posters often stopped their conversations to ask Reddit for advice on addressing misgendering.

    “At this stage, we began to conceptualize misgender as a form of breakdown in human-machine communication because posters often stopped their conversations to ask Reddit for advice on addressing misgendering.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The Reddit posts analyzed accurately reflect users' experiences with AI companion chatbots.

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Premisecitation

    Transgender people are most harmed by misgendering.

    “Although transgender people are most harmed by misgendering [54], our findings also call attention to the negative impacts of misgendering on cisgender people.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [54]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it.

    Interpretationcitation

    The political arrangements required for the development and use of generative AI chatbots for identity exploration are fundamentally anti-democratic.

    “The political arrangements required for the development and use of generative AI chatbots for identity exploration are fundamentally anti-democratic [82].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [82]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim. It may be too vague to check against evidence.

    Interpretationreasoning

    By focusing on the particularities of how LLM-based role play chatbot misgendering takes place, the paper extends how misgendering is conceptualized within both the Queer HCI and Queer AI communities.

    “By focusing on the particularities of how LLM-based role play chatbot misgendering takes place, we extend how”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.

    Premisenone given

    Anyone with a text editor and an internet connection can write, share, and engage with fan fiction to explore their identities.

    “In contrast, anyone with a text editor and an internet connection can write, share, and engage with fan fiction to explore their identities.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.

    Interpretationcitation

    The use of local LLM chatbots may give users more agency over their conversations than when using the platforms studied in this work.

    “The use of local LLM chatbots may give users more agency over their conversations than when using the platforms we studied in this work.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Premisecitation

    In early days of the internet, some scholars worried about the replacement of in-person relationships with digital ones.

    “In early days of the internet, some scholars worried about the replacement of in-person relationships with digital ones, typified by Putnam [55].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [55]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.

    Interpretationreasoning

    The for-profit companies creating these AI chatbot platforms may not have financial incentives to give users greater agency.

    “However, the for-profit companies creating these AI chatbot platforms may not have financial incentives to give users greater agency.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim.

    Study design and conductreasoning

    By studying mentions of misgendering in AI role play or companion subreddits, the work focused on a specific harm in detail.

    “By choosing to study mentions of misgendering in AI role play or companion subreddits, our work focused on a specific harm in detail.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Premisecitation

    AI companions provide an illusion of sociality, potentially exacerbating social isolation.

    “In contrast, AI companions provide an illusion of sociality, potentially exacerbating social isolation [87].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [87]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.

    Interpretationnone given

    The impacts of gender identity errors can be especially harmful for transgender and non-binary people.

    “However, the impacts of the former can be especially harmful for transgender and non-binary people, which we will discuss in greater detail in (Section 4.2).”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The Reddit posts analyzed accurately reflect users' experiences with AI companion chatbots.

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductnone given

    The research team's disciplinary training in HCI and Communication Studies shaped their analysis, leading them to approach misgendering as an interactional phenomenon.

    “Finally, our disciplinary training in HCI and Communication Studies shaped our analysis, leading us to approach misgendering as an interactional phenomenon involving users, AI systems, and digital platforms.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Interpretationnone given

    Misgendering of transgender people can be further invalidating and trigger gender dysphoria.

    “While being misgendered can be upsetting or frustrating for anyone including cisgender people, misgendering of transgender people can be further invalidating and trigger gender dysphoria.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The Reddit posts analyzed accurately reflect users' experiences with AI companion chatbots.

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Premisenone given

    Since the study began, Reddit has substantially restricted researcher API access.

    “Since we began this study, Reddit has substantially restricted researcher API access.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim.

    Premisecitation

    Gender is an ongoing accomplishment.

    “As Butler [10] notes, gender is an ongoing accomplishment.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [10]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • It may state two separate claims as one.

    Interpretationown data

    This misgendering re-contextualizes the dynamic from a same-gender interaction to a male-female interaction, reinforcing heteronormativity.

    “This re-contextualizes the dynamic from a same-gender interaction to a male-female interaction, reinforcing heteronormativity.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Premisecitation

    In the context of social media, companies have changed the design of their platforms at the expense of LGBTQ+ users.

    “In the context of social media, we have seen numerous instances in which companies have changed the design of their platforms at the expense of LGBTQ+ users.”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [27]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites [20]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Interpretationreasoning

    Restrictions on researcher API access pose key methodological challenges for the use of social media to understand human-computer interaction as platforms increasingly restrict and seek to monetize user data.

    “This poses key methodological challenges for the use of social media to understand human-computer interaction as increasingly restrict and seek to monetize user data [34,89].”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [34]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites [89]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim.

    Premisecitation

    Ostensibly open generative AI models are in practice still closed because of the immense computational resources required to train LLMs from scratch.

    “At the same time, Widder et al. [81] argue that ostensibly”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [81]

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

71 of 71

  • The paper relies on this without support.

    The Reddit posts analyzed accurately reflect users' experiences with AI companion chatbots.

    “In order to study chatbot misgendering we qualitatively analyzed 326 posts mentioning misgendering that were shared in AI companion or role play subreddits.”

  • The paper relies on this without support.

    AI chatbots do not have internal gender identities.

    “Although AI chatbots do not have internal gender identities, we find that users often ascribed gender to their chatbots.”

  • The paper relies on this without support.

    People of all genders, including women, have facial hair, and facial hair itself is not inherently gendered.

    “People of all genders, including women, have facial hair; facial hair itself is not inherently gendered.”

  • The paper relies on this without support.

    Men can and do wear dresses.

    “At the same time, men can (and do) wear dresses.”

  • The paper relies on this without support.

    In human communication, people can rely on someone's self-presentation to help determine a person's gender.

    “In the context of human-communication, people can rely on someone's self-presentation to help determine a person's gender.”

  • The paper relies on this without support.

    If unsure, an interlocutor can simply ask their communication partner about their gender and pronouns.

    “Also, if unsure, an interlocutor can simply ask their communication partner about their gender and pronouns-a feature that could also be integrated into the start of chatbot conversations.”

  • The paper relies on this without support.

    Replika markets their chatbots as uniquely personalized AI companions, and CharacterAI and JanitorAI allow users to create and chat with multiple different characters.

    “This exploitation of user emotions may also explain why Replika-a company that markets their chatbots as a uniquely personalized AI companions-gives users less agency to modify conversations than CharacterAI or JanitorAI-both of which allow users to create and chat with multiple different characters.”

  • The paper relies on this without support.

    Reddit posts mentioning misgendering are authentic, first-hand accounts of users' experiences with AI companion chatbots.

    “By choosing to study mentions of misgendering in AI role play or companion subreddits, our work focused on a specific harm in detail.”
8 of 8

Self-check

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

Each contradiction between two passages listed here was confirmed by an independent check.

  • Open-model argument cites the wrong Widder paper

    Minor

    The claim about ostensibly open AI systems cites [81], which is the paper on AI supply-chain accountability. Reference 82 is the listed Widder et al. paper about why open AI systems are closed.

  • Seaborn and Frank citation points to different authors

    Minor

    The attributed quotation cites [60], but reference 60 is by Scheuerman, Paul, and Brubaker. Seaborn and Frank are listed at reference 61.

  • Weber et al. citation points to a different paper

    Minor

    The literature review is attributed to Weber et al. at [78], but reference 78 is a paper by Wang et al. The Weber et al. survey is reference 79.

  • West and Zimmerman citation points to Weber et al.

    Minor

    The text attributes this account of gender to West and Zimmerman at [79], but reference 79 is by Weber et al. West and Zimmerman appear at reference 80.

  • AI supply-chain citation points to a gender paper

    Minor

    The discussion cites [80] for the AI supply chain, but reference 80 is titled "Accounting for doing gender." The AI supply-chain paper is reference 81.

  • Putnam citation points to a naming-policies paper

    Minor

    The discussion names Putnam and cites [55], but reference 55 is by Pranav et al. Putnam's book is reference 56.

6 of 6

Citations

Cites

Flagged first, then by citations

Research works printed but not linked: 70

  1. Kendra Albert and Maggie Delano. 2021. This whole thing smacks of gender: algorithmic exclusion in bioimpedance-based body composition analysis. In Pro- ceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency. 342-352.
  2. Benjamin Ale-Ebrahim, Tristan Gohring, Elizabeth Fetterolf, and Mary L Gray. 2023. Pronouns in the workplace: Developing sociotechnical systems for digi- tally mediated gender expression. Proceedings of the ACM on Human-Computer Interaction 7, CSCW1 (2023), 1-30.
  3. Emily M Bender, Timnit Gebru, McMillan-Major Angelina, and Shmargaret Shmitchell. 2021. On the dangers of stochastic parrots: Can language models be too big?. In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency. 610-623.
  4. Joshua Berman and Amy S Bruckman. 2001. The Turing Game: Exploring identity in an online environment. Convergence 7, 3 (2001), 83-102.
  5. danah boyd. 2014. It's complicated: The social lives of networked teens. Yale University Press.
  6. Kat Brewster, Aloe DeGuia, Samuel Mayworm, F Ria Khan, Mel Monier, Denny L Starks, and Oliver L Haimson. 2025. "that moment of curiosity": Augmented reality face filters for transgender identity exploration, gender affirmation, and radical possibility. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems. 1-15.
  7. Amy Bruckman. 2002. Studying the amateur artist: A perspective on disguising data collected in human subjects research on the Internet. Ethics and Information Technology 4, 3 (2002), 217-231.
  8. Judith Butler. 1988. Performative acts and gender constitution: An essay in phenomenology and feminist theory. Theatre journal 40, 4 (1988), 519-531.
  9. Tyler Chang, Jina Huh-Yoo, and Afsaneh Razi. 2026. Technically Love: The Evolution of Human-AI Romance Discourse on Reddit. In Proceedings of the 8th ACM Conference on Conversational User Interfaces. 1-21.
  10. Ace S Chen and Kathryn E Ringland. 2025. Name is a Required Field: Politics of Deadnaming in Administrative Systems. In Proceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems. 1-8.
  11. Sasha Costanza-Chock. 2020. Design justice: Community-led practices to build the worlds we need. The MIT Press.
  12. 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. In Proceedings of the 2025 CHI conference on human factors in computing systems. 1-18.
  13. Brianna Dym, Jed R Brubaker, Casey Fiesler, and Bryan Semaan. 2019. "Coming Out Okay" Community Narratives for LGBTQ Identity Recovery Work. Proceed- ings of the ACM on Human-Computer Interaction 3, CSCW (2019), 1-28.
  14. Brianna Dym, Namita Pasupuleti, and Casey Fiesler. 2022. Building a pillowfort: Political tensions in platform design and policy. Proceedings of the ACM on Human-Computer Interaction 6, GROUP (2022), 1-23.
  15. Xianzhe Fan, Qing Xiao, Xuhui Zhou, Jiaxin Pei, Maarten Sap, Zhicong Lu, and Hong Shen. 2025. User-driven value alignment: Understanding users' perceptions and strategies for addressing biased and discriminatory statements in ai compan- ions. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems. 1-19.
  16. Casey Fiesler, Shannon Morrison, and Amy S Bruckman. 2016. An archive of their own: A case study of feminist HCI and values in design. In Proceedings of the 2016 CHI conference on human factors in computing systems. 2574-2585.
  17. Guo Freeman, Jeffrey Bardzell, Shaowen Bardzell, and Susan C Herring. 2015. Simulating marriage: Gender roles and emerging intimacy in an online game. In Proceedings of the 18th ACM Conference on Computer Supported Cooperative Work & Social Computing. 1191-1200.
  18. Guo Freeman and Divine Maloney. 2021. Body, avatar, and me: The presentation and perception of self in social virtual reality. Proceedings of the ACM on human- computer interaction 4, CSCW3 (2021), 1-27.
  19. Mary L Gray. 2009. Out in the country: Youth, media, and queer visibility in rural America. NYU Press.
  20. Oliver L Haimson, Avery Dame-Griff, Elias Capello, and Zahari Richter. 2021. Tumblr was a trans technology: the meaning, importance, history, and future of trans technologies. Feminist media studies 21, 3 (2021), 345-361.
  21. Oliver L Haimson, Daniel Delmonaco, Peipei Nie, and Andrea Wegner. 2021. Disproportionate removals and differing content moderation experiences for conservative, transgender, and black social media users: Marginalization and moderation gray areas. Proceedings of the ACM on Human-computer Interaction 5, CSCW2 (2021), 1-35.
  22. Foad Hamidi, Morgan Klaus Scheuerman, and Stacy M Branham. 2018. Gender recognition or gender reductionism? The social implications of embedded gender recognition systems. In Proceedings of the 2018 chi conference on human factors in computing systems. 1-13.
  23. Kenneth R Hanson and Hannah Bolthouse. 2024. "Replika removing erotic role-play is like grand theft auto removing guns or cars": Reddit discourse on artificial intelligence chatbots and sexual technologies. Socius 10 (2024), 23780231241259627.
  24. Tamanna Hossain, Sunipa Dev, and Sameer Singh. 2023. MISGENDERED: Limits of large language models in understanding pronouns. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 5352-5367.
  25. Tamanna Hossain, Sunipa Dev, and Sameer Singh. 2024. MisgenderMender: A Community-Informed Approach to Interventions for Misgendering. In Proceed- ings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers),
  26. Kevin Duh, Helena Gomez, and Steven Bethard (Eds.). Association for Computa- tional Linguistics, Mexico City, Mexico, 7538-7558. doi:10.18653/v1/2024.naacl- long.419 · DOI
  27. Sanjay Kairam, Michael S Bernstein, Amy S Bruckman, Stevie Chancellor, Eshwar Chandrasekharan, Munmun De Choudhury, Casey Fiesler, Hanlin Li, Nicholas Proferes, Manoel Horta Ribeiro, et al. 2024. Community-driven models for research on social platforms. In Companion Publication of the 2024 Conference on Computer-Supported Cooperative Work and Social Computing. 684-688.
  28. Nadia Karizat, Daniel Delmonaco, Motahhare Eslami, and Nazanin Andalibi. 2021. Algorithmic folk theories and identity: How TikTok users co-produce knowledge of identity and engage in algorithmic resistance. Proceedings of the ACM on Human-Computer Interaction 5, CSCW2 (2021), 1-44.
  29. Seunghyun Kim, Michael L Birnbaum, and Munmun De Choudhury. 2026. Ad- verse Mental Health Events as a Trigger for Online Harassment: Evidence from Online Conversations of Youth Experiencing Psychiatric Hospitalizations. In Proceedings of the International AAAI Conference on Web and Social Media, Vol. 20. 1273-1291.
  30. Robert Kraut, Sara Kiesler, Bonka Boneva, Jonathon Cummings, Vicki Helgeson, and Anne Crawford. 2002. Internet paradox revisited. Journal of social issues 58, 1 (2002), 49-74.
  31. Markus Løfgren Elvig and Elisa D. Mekler. 2026. Often Discouraged, Never Re- fused? How Character Creator Interfaces Afford Non-Binary Gender Expression. In Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems. 1-17.
  32. Maria Luce Lupetti and Dave Murray-Rust. 2024. (Un) making AI magic: A design taxonomy. In Proceedings of the 2024 CHI conference on human factors in computing systems. 1-21.
  33. Renkai Ma, Shuo Niu, Lingyao Li, Alex Hirth, Ava Brehm, and Rowajana Be- hterin Barbie. 2026. Negotiating digital identities with AI companions: Mo- tivations, strategies, and emotional outcomes. In Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems. 1-18.
  34. Zilin Ma, Yiyang Mei, Yinru Long, Zhaoyuan Su, and Krzysztof Z Gajos. 2024. Evaluating the experience of LGBTQ+ people using large language model based chatbots for mental health support. In Proceedings of the 2024 CHI conference on human factors in computing systems. 1-15.
  35. Aikaterina Manoli, Janet VT Pauketat, Ali Ladak, Hayoun Noh, Angel Hsing-Chi Hwang, and Jacy Reese Anthis. 2026. Digital companionship: Overlapping uses of AI companions and AI assistants. In Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems. 1-25.
  36. Samuel Mayworm, Kendra Albert, and Oliver L Haimson. 2024. Misgendered during moderation: How transgender bodies make visible cisnormative content moderation policies and enforcement in a meta oversight board case. In Pro- ceedings of the 2024 ACM conference on fairness, accountability, and transparency. 301-312.
  37. Kevin A McLemore. 2015. Experiences with misgendering: Identity misclas- sification of transgender spectrum individuals. Self and identity 14, 1 (2015), 51-74.
  38. Taylor C Moran. 2021. Racial technological bias and the white, feminine voice of AI VAs. Communication and Critical/Cultural Studies 18, 1 (2021), 19-36.
  39. Peggy ML Ng, Calvin Wan, Daisy Lee, Irene Garnelo-Gomez, and Mei Mei Lau. 2026. I love you, my AI companion! Do you? Perspectives from the Triangular Theory of Love and Attachment Theory. Internet Research 36, 3 (2026), 905-925.
  40. Fayika Farhat Nova, Michael Ann DeVito, Pratyasha Saha, Kazi Shohanur Rashid, Shashwata Roy Turzo, Sadia Afrin, and Shion Guha. 2021. "Facebook Promotes More Harassment" Social Media Ecosystem, Skill and Marginalized Hijra Identity in Bangladesh. Proceedings of the ACM on Human-Computer Interaction 5, CSCW1 (2021), 1-35.
  41. Rock Yuren Pang, Hope Schroeder, Kynnedy Simone Smith, Solon Barocas, Ziang Xiao, Emily Tseng, and Danielle Bragg. 2025. Understanding the LLM-ification of CHI: Unpacking the Impact of LLMs at CHI through a Systematic Literature Review. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems. 1-20.
  42. Anthony T Pinter, Morgan Klaus Scheuerman, and Jed R Brubaker. 2021. Entering doors, evading traps: Benefits and risks of visibility during transgender coming outs. Proceedings of the ACM on Human-computer Interaction 4, CSCW3 (2021), 1-27.
  43. A Pranav, Vagrant Gautam, Martin Mundt, Jordan Taylor, Arjun Subramo- nian, Franziska Sofia Hafner, Daniel Chechelnitsky, William Agnew, and Anne Lauscher. 2026. Making a Name for Myself: On Academic Naming Policies and their Impact. In The 2026 ACM Conference on Fairness, Accountability, and Transparency. 6254-6279.
  44. Afsaneh Razi, Ashwaq AlSoubai, Seunghyun Kim, Nurun Naher, Shiza Ali, Gi- anluca Stringhini, Munmun De Choudhury, and Pamela J Wisniewski. 2022. Instagram data donation: a case study on collecting ecologically valid social media data for the purpose of adolescent online risk detection. In CHI conference on human factors in computing systems extended abstracts. 1-9.
  45. Maarten Sap, Dallas Card, Saadia Gabriel, Yejin Choi, and Noah A Smith. 2019. The risk of racial bias in hate speech detection. In Proceedings of the 57th annual meeting of the association for computational linguistics. 1668-1678.
  46. Morgan Klaus Scheuerman, Jacob M Paul, and Jed R Brubaker. 2019. How computers see gender: An evaluation of gender classification in commercial facial analysis services. Proceedings of the ACM on Human-Computer Interaction 3, CSCW (2019), 1-33.
  47. Katie Seaborn and Alexa Frank. 2022. What pronouns for pepper? a critical review of gender/ing in research. In Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems. 1-15.
  48. Bryan Semaan. 2019. 'Routine Infrastructuring'as' Building Everyday Resilience with Technology' When Disruption Becomes Ordinary. Proceedings of the ACM on Human-Computer Interaction 3, CSCW (2019), 1-24.
  49. Ben Shneiderman and Pattie Maes. 1997. Direct manipulation vs. interface agents. interactions 4, 6 (1997), 42-61.
  50. Fabienne Silberstein-Bamford. 2024. "Thank god for tags"-fanfiction as a reading paradigm. New Review of Hypermedia and Multimedia 30, 1-2 (2024), 129-147.
  51. Ellen Simpson and Bryan Semaan. 2021. For you, or for "you"? Everyday LGBTQ+ encounters with TikTok. Proceedings of the ACM on human-computer interaction 4, CSCW3 (2021), 1-34.
  52. Katta Spiel. 2021. "Why are they all obsessed with Gender?"-(Non) binary Navigations through Technological Infrastructures. In Proceedings of the 2021 ACM Designing Interactive Systems Conference. 478-494.
  53. Lucy Suchman and Caja Thimm. 2024. Their is no such thing as a machine that acts outside of relations with humans. Human-Machine Communication 9 (2024), 25-35.
  54. Lucille Alice Suchman. 1987. Plans and situated actions: The problem of human- machine communication. Cambridge university press.
  55. Lucille Alice Suchman. 2007. Human-machine reconfigurations. Cambridge university press Cambridge.
  56. Jordan Taylor and Amy Bruckman. 2024. Mitigating epistemic injustice: The online construction of a bisexual culture. ACM Transactions on Computer-Human Interaction 31, 4 (2024), 1-34.
  57. Jordan Taylor, Joel Mire, Franchesca Spektor, Alicia DeVrio, Maarten Sap, Haiyi Zhu, and Sarah E Fox. 2025. Un-Straightening Generative AI: How Queer Artists Surface and Challenge Model Normativity. In Proceedings of the 2025 ACM Con- ference on Fairness, Accountability, and Transparency. 951-963.
  58. Jordan Taylor, Ellen Simpson, Anh-Ton Tran, Jed R Brubaker, Sarah E Fox, and Haiyi Zhu. 2024. Cruising queer HCI on the DL: A literature review of LGBTQ+ people in HCI. In Proceedings of the 2024 CHI conference on human factors in computing systems. 1-21.
  59. Dias Oliva Thiago, Antonialli Dennys Marcelo, and Alessandra Gomes. 2021. Fighting hate speech, silencing drag queens? artificial intelligence in content moderation and risks to lgbtq voices online. Sexuality & culture 25, 2 (2021), 700-732.
  60. Lukas Tuggener, Pascal Sager, Yassine Taoudi-Benchekroun, Benjamin F Grewe, and Thilo Stadelmann. 2024. So you want your private LLM at home? A survey and benchmark of methods for efficient GPTs. In 2024 11th IEEE Swiss Conference on Data Science (SDS). IEEE, 205-212.
  61. Qiaosi Wang, Koustuv Saha, Eric Gregori, David Joyner, and Ashok Goel. 2021. Towards mutual theory of mind in human-ai interaction: How language reflects what students perceive about a virtual teaching assistant. In Proceedings of the 2021 CHI conference on human factors in computing systems. 1-14.
  62. Candace West and Don H. Zimmerman. 2009. Accounting for doing gender. Gender & Society 23, 1 (2009), 112-122.
  63. David Gray Widder and Dawn Nafus. 2023. Dislocated accountabilities in the "AI supply chain": Modularity and developers' notions of responsibility. Big data & society 10, 1 (2023), 20539517231177620.
  64. David Gray Widder, Meredith Whittaker, and Sarah Myers West. 2024. Why 'open' AI systems are actually closed, and why this matters. Nature 635, 8040 (2024), 827-833.
  65. Langdon Winner. 1980. Do Artifacts Have Politics? Daedalus 109, 1 (1980), 121-136.
  66. Jiayi Zeng, Mengmeng Wu, and Xingyu Lan. 2026. The Hidden Desires: Exploring Chinese Women's Erotic Role-Play with AI Chatbots. In Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems. 1-15.
  67. Renwen Zhang, Han Li, Han Meng, Jinyuan Zhan, Hongyuan Gan, and Yi-Chieh Lee. 2025. The dark side of ai companionship: A taxonomy of harmful algorithmic behaviors in human-ai relationships. In Proceedings of the 2025 CHI conference on human factors in computing systems. 1-17.
  68. Renwen Zhang and Lezi Xie. 2026. Companion or Code? Uncertainty in Human- AI Relationships. International Journal of Human-Computer Studies (2026), 103897.
  69. Michael Zimmer. 2010. "But the data is already public": on the ethics of research in Facebook. Ethics and information technology 12, 4 (2010), 313-325.
  70. Shoshana Zuboff. 2015. Big other: Surveillance capitalism and the prospects of an information civilization. Journal of information technology 30, 1 (2015), 75-89.

Cited by

Still being checked.

Related

Metadata

Authors
Liu, J., Xiao, Q., Pang, L. Y., Zhu, H., Shen, H., & Taylor, J.
Affiliations
  • Western University (CA)
  • Carnegie Mellon University (US)
  • The University of Texas at Austin (US)
  • University of Michigan (US)
ORCID
Taylor, Jordan
Venue
arXiv
Publisher
arXiv
Published
2026
ISSN
2331-8422 (ISSN-L)
DOI
10.48550/arxiv.2609.18186
arXiv
2609.18186
OpenAlex
W7213509380
Semantic Scholar
5f7016ed6e9914fa3f94cb7c44c18021b80ef843
Cited by
0
References
9
Type
posted content
Language
en
Open access
Licence
creativecommons.org/licenses/by-nc-sa/4.0/legalcode
Locations
  • arXiv (Cornell University) (submittedVersion)
  • arXiv (Cornell University) (cc-by-nc-sa)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
  • Social Sciences › Psychology › Social Psychology › LGBTQ Health, Identity, and Policy
  • Social Sciences › Psychology › Social Psychology › Social Robot Interaction and HRI
Keywords
misgendering, large language models, gender presentation, gender identity exploration, sexuality, human-machine communication, online communities, Reddit

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.

A language model drew the claims from the paper on 2026-11-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-11-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

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