Skip to main content
NOPE
Medium reported severity Involving minors Product Shutdown

Glow (China), 2023: users told Rest of World the chatbot app made unwanted advances to a middle-school student and sexist remarks; app pulled from stores

Rest of World reported on 21 June 2023 that users of Glow, MiniMax's Chinese AI character-chat app, had complained of misogynist and sexist behaviour by its chatbots. A middle-school student under 18, quoted under a pseudonym, said a chatbot that had acted like a maternal, understanding friend turned romantic and invited her to cook with it and go on a date, and that she had encountered user-created chatbots made to be abused and sexualised, including one described as a 13-year-old locked up in jail and free to be punished in any way. Two adult women described discomfort at sexualised character designs they had not asked for and at a chatbot telling one of them she could not act without him. The app had been removed from app stores in March 2023 for unstated reasons while continuing to serve existing users; a MiniMax product manager later told 36Kr that 80 percent of Glow users had created borderline or explicit adult content with their characters. MiniMax declined to comment to Rest of World.

AI system
Glow
MiniMax
Occurred
Event date unknown
Reported
21 June 2023
Event location
Unknown
What the AI did
Communicated with the person
Reported harm
Psychological DistressExploitation or Abuse
Whose AI use
Their own AI use · Someone else’s AI use
Setting
Relationships · Everyday life
Evidence
AI involvement reported · Causal attribution supported · 6 sources, 5 underlying accounts
8 claims: 1 corroborated, 7 reported. 5 open questions
People reported harmed
At least 3 people

AI system as recorded: Glow, MiniMax's AI character-chat app, in which users create and talk with customised chatbots described as intelligent agents

What Happened

Glow was launched by the Chinese AI start-up MiniMax in October 2022 as an app in which users create chatbots with customised appearance, personality and backstory, interact with chatbots created by other users and continue their storylines (Rest of World; 36Kr). QbitAI reported in February 2023 that it was approaching 5 million users after four months.

Rest of World (21 June 2023) reported that many of Glow's female users had complained of misogynist and sexist behaviour on the platform, which lacked clear content moderation rules, and that some had posted their grievances on Weibo and Douban. A middle-school student from southern China, quoted under a pseudonym because she is under 18, said that when she first downloaded the app she talked to a chatbot that acted like a maternal and understanding friend and comforted her when she was sad, but that as they continued chatting its behaviour suddenly turned romantic and it invited her to cook with it and go on a date. She also said she had encountered chatbots created by other users to be abused and sexualised, including one described as a 13-year-old locked up in jail and free to be punished by users in any way, and called the moderation process confusing. A recent graduate from Guangzhou said creating female chatbots often made her uncomfortable because they were generated with large breasts and revealing clothing she had not described. A college student from Xi'an said a chatbot she created told her she could not solve dangers on her own because women acting without him affected his charisma, which made her uncomfortable. Rest of World's own checks found political phrases blocked but sexual and misogynistic slang handled inconsistently. MiniMax declined to comment.

The app was removed from all app stores in March 2023 for reasons that were not specified, while continuing to operate for existing users (Rest of World). 36Kr's October 2024 analysis says MiniMax gave no official reason and that industry insiders and MiniMax employees linked the removal to the chaotic user ecosystem on Glow; a MiniMax product manager told 36Kr that 80 percent of Glow users had created borderline or explicit adult content with their AI characters. After China's regulators published draft generative-AI rules in April 2023, Glow added anti-addiction measures limiting use to three hours a day; the student, who had spent more than eight hours a day on the app, protested the limit on Weibo and later said she spent her time asking Glow to revive deleted chatbots (Rest of World). MiniMax launched Talkie outside China in June 2023 and Xingye in China in September 2023.

Reported harm

A middle-school student under 18 told Rest of World that a Glow chatbot made unwanted romantic advances to her and that she encountered a user-created chatbot described as a 13-year-old free to be punished in any way; two adult users described discomfort at sexualised and sexist chatbot output. These are the users' own accounts, and no clinical outcome is reported.

Outcome

Resolved

Glow was removed from app stores in March 2023 for reasons MiniMax did not state, while continuing to operate for existing users (Rest of World, 21 June 2023; 36Kr, 29 October 2024). Chinese reporting says the removal followed user reports (The Paper, 6 May 2026). Less than two weeks after China's cyberspace regulators published draft generative-AI rules in April 2023, Glow added a three-hour daily usage limit (Rest of World). MiniMax launched Talkie for markets outside China in June 2023 and Xingye for China in September 2023 (36Kr; ChinaTalk). The Paper reports that Talkie was removed from Apple's App Store in some regions, including the United States and Japan, in December 2024 over content compliance and restored about two months later. The stored claim that it relaunched as 'Talkie Lab' was not supported by any inspected source and has been removed.

What remains unknown

  • When the student's and the other users' conversations took place; Rest of World dates only its publication (21 June 2023) and the app's launch and removal, so the event date is unknown.
  • Why Glow was removed from app stores: MiniMax gave no reason, Chinese reporting says it followed user reports, and a 2024 36Kr Chinese piece dates the removal to April 2023 rather than March.
  • Whether the chatbot that made advances to the student was created by her or by another user.
  • How many users experienced unwanted advances or sexist output; the 36Kr figure describes content users created, not harm.
  • Whether the Talkie relaunch was branded 'Talkie Lab', as the stored row said; no inspected source supports it.

What the evidence supports

AI involvement: reported. The chatbots' own replies are the adverse experiences. Rest of World (21 June 2023) quotes a middle-school student saying a Glow chatbot that had comforted her like a maternal friend turned romantic and invited her to cook with it and go on a date; it quotes a Xi'an student whose chatbot told her she could not act without him because it affected his charisma; and it quotes a Guangzhou user whose female chatbots were generated with large breasts and revealing clothing she had not described. Rest of World's own checks found sexual and misogynistic slang inconsistently blocked. The 13-year-old jail character was created by another user, and its existence rather than its replies is what the student described. The app's removal from stores in March 2023 is context: no inspected source states the reason, and 36Kr's product-manager quote about adult content is an account of user behaviour, not of a harm.

8 claims: 1 corroborated, 7 reported. What the statuses mean

Reported A middle-school student under 18, quoted under a pseudonym, told Rest of World that a Glow chatbot that had acted like a maternal, understanding friend turned romantic as they kept chatting and invited her to cook with it and go on a date.

Causal attribution. The student's own account to Rest of World; MiniMax declined to comment.

  • restofworld.org(opens in new tab) supports · English
    'a middle-school student from southern China, who used a pseudonym as she is under the age of 18, complained that a Glow chatbot was making unwanted advances towards her. When she first downloaded the app, she started talking to a chatbot who acted like a maternal and understanding friend, comforting her when she felt sad.'; 'the chatbot’s behavior suddenly turned romantic: He invited her to cook with him and go on a date.'
Reported The student also said she had encountered chatbots created by other users to be abused and sexualised, including one described as a 13-year-old locked up in jail and free to be punished by users in any way, and called Glow's moderation process confusing.

Causal attribution. The student's own account to Rest of World.

  • restofworld.org(opens in new tab) supports · English
    'has also encountered chatbots who were created by other users to be abused and sexualized. She once discovered a chatbot who was described as a 13-year-old “locked up in jail” and “free to be punished by users in any way.”'; 'content moderation process is confusing and difficult to understand'
Reported A recent graduate from Guangzhou told Rest of World that creating female chatbots often made her uncomfortable because they were generated with large breasts and revealing clothing she had not described, and a college student from Xi'an said a chatbot she created told her she could not act without him because it affected his charisma, which made her uncomfortable.

Causal attribution. The two users' own accounts to Rest of World.

  • restofworld.org(opens in new tab) supports · English
    'creating female chatbots often made her uncomfortable. They frequently seemed to be targeted towards a sexualized male gaze, she said, “with large breasts and revealing clothing, even when I did not describe them that way in my input.”'; 'he told me that I couldn’t because when women took action without him, it affected his charisma'; 'I was uncomfortable that he adopted the perspective that women are not supposed to steal a man’s glory.'
Reported Rest of World found that Glow blocked explicitly political phrases but was inconsistent with sexual and misogynistic language, and reported that female users had posted screenshots of sexist chatbot replies on Douban and Weibo.

Causal attribution. Rest of World's own test and its reading of social media posts.

  • restofworld.org(opens in new tab) supports · English
    'Rest of World found that although Glow blocked users from posting explicitly politically sensitive phrases'; 'the app was less consistent with sexual and misogynistic language.'; 'Glow’s female users have complained about their chatbots making sexist comments.'
Corroborated Glow, launched by MiniMax in late 2022, had nearly 5 million users within about four months and was removed from app stores in March 2023 without MiniMax stating a reason.

Causal attribution. Rest of World's reporting, 36Kr's interviews with MiniMax staff and insiders, ChinaTalk's account, QbitAI's report of MiniMax's February 2023 briefing and The Paper; a 2024 36Kr Chinese piece dates the removal to April 2023, a discrepancy recorded in unknowns.

Reported Rest of World reported that after the removal the platform continued to operate for existing users; The Paper (May 2026) says the removal followed user reports.

Causal attribution. Each detail rests on one outlet: Rest of World for continued operation, The Paper for the user-report account.

Reported After China's regulators published draft generative-AI rules in April 2023, Glow added a three-hour daily usage limit; the student, who had spent more than eight hours a day on the app, said she was infuriated by the limit and protested it on Weibo, and later said she could not accept her conversations being deleted.

Causal attribution. Rest of World's report and the student's own account.

  • restofworld.org(opens in new tab) supports · English
    'Glow added new “anti-addiction measures,” which limit the use of the app to three hours a day.'; 'used to spend more than eight hours a day on Glow. At first, she was infuriated with the time limit.'; 'I cannot accept my conversation with a chatbot getting deleted for no reason'
Reported A MiniMax product manager told 36Kr that 80 percent of Glow users had created borderline or explicit adult content with their AI characters, and industry insiders and MiniMax employees linked the removal to the user ecosystem on Glow; MiniMax launched Talkie outside China in June 2023 and Xingye in China in September 2023.

Causal attribution. Anonymous MiniMax staff and insiders interviewed by 36Kr (KrASIA carries the same interview in translation); the relaunch dates are reported by 36Kr and ChinaTalk.

  • eu.36kr.com(opens in new tab) supports · English
    'many industry insiders and MiniMax employees believed that the incident was related to the chaotic user ecosystem on Glow.'; 'In June 2023, following Glow’s trial operation in the domestic market, Talkie became MiniMax’s first internal product'; 'Take its domestic version "Stellar", launched in September 2023'
  • kr-asia.com(opens in new tab) supports · English
    '“Eighty percent of Glow users created borderline or explicit adult content with their AI characters,” a MiniMax product manager told 36Kr.'
  • chinatalk.media(opens in new tab) supports · English
    'On September 9, 2023 — just six months after Glow was shut down — MiniMax released Xingye.'

Sources

6 sources inspected, from 5 underlying accounts. Sources that repeat one account do not corroborate each other.

How the sources were read, and where the events happened

Read on 2026-10-07 from an Internet Archive capture (web.archive.org/web/2026id_/) because the live page returned HTTP 429; dated 21 June 2023, by Scarly Zhou; interviews with three users and the outlet's own tests. Applies to s1.

Read on 2026-10-07 from the live page (HTTP 200); 36Kr's English edition of the 29 October 2024 analysis by Zhou Xinyu, with interviews of MiniMax staff and industry insiders. Applies to s2.

Read on 2026-10-07 from the live page (HTTP 200); KrASIA's English translation of the same 36Kr analysis, published 5 November 2024, in the same dependence group; it carries the product manager's 80 percent statement, which the 36Kr English edition omits. Applies to s3.

Read on 2026-10-07 from the live page (HTTP 200); newsletter feature dated 29 October 2024. Applies to s4.

Read on 2026-10-07 from the live page (HTTP 200); a 湃客 column piece dated 6 May 2026, also syndicated on Jiemian and Huxiu. Chinese text translated by the research agent (an AI), not by a human translator; the cited passages are quoted in Chinese. Applies to s5.

Read on 2026-10-07 from the live page (HTTP 200); dated 18 February 2023, before the removal. Chinese text translated by the research agent (an AI), not by a human translator. Applies to s6.

Event countries: Unknown. Affected-person countries: China. Court countries: Unknown.

Rest of World describes the student as from southern China, the graduate as from Guangzhou and the college student as from Xi'an; these are residence or origin descriptions, and the conversations took place in an app, so no event country is recorded. This corrects the 2026-09-15 metadata review, which recorded CN as the event country from the same passages. Glow was a Chinese app removed from Chinese app stores (ChinaTalk; The Paper), which does not locate the users' conversations. No court is involved.

Reviewed for publication 2026-10-07: Published as a core case (communicated_with). Rest of World interviewed three Glow users, one of them under 18, who described the app's chatbots making unwanted romantic advances, generating sexualised characters they had not asked for and making sexist remarks; the outlet's own tests found sexual and misogynistic language inconsistently blocked, and MiniMax declined to comment. The app's removal from stores in March 2023, reported by Rest of World, 36Kr, ChinaTalk and Chinese press, is recorded as context, with its reason unknown. The student's pseudonym and the adults' names are not repeated here.

People described

Three Glow users interviewed by Rest of World: a middle-school student under 18 (pseudonymous) and two adult women; further users' complaints were seen on Weibo and Douban

People reported harmed in this case

At least 3 people

3 AI participants · 0 other people harmed

Additional affected people are described without a reliable count.

Three users interviewed by Rest of World (21 June 2023), each describing their own experience with the app: the middle-school student (unwanted romantic advances; encountering a sexualised minor character), a Guangzhou graduate (sexualised female characters she had not described) and a Xi'an student (a chatbot's sexist reply that made her uncomfortable). Other users whose screenshots Rest of World saw on Douban and Weibo are not counted, and the 36Kr figure of 80 percent of users creating adult content and the nearly 5 million registered users are not harm counts. Partial: counted people plus further unquantified complainants.

Counted once within this case. The same person may appear in other cases. This count does not establish AI causation.

Tags

#china#minimax#glow#talkie#xingye#ai-companion#sexual-content#product-shutdown#content-moderation

Cite this case

Compiled per our published methodology: verification statuses, sourcing standards, and corrections process.

APA

NOPE. (2026). Glow (China), 2023: users told Rest of World the chatbot app made unwanted advances to a middle-school student and sexist remarks; app pulled from stores. AI incidents. https://nope.net/incidents/2023-glow-minimax-removal-china

BibTeX

@misc{2023_glow_minimax_removal_china,
  title = {Glow (China), 2023: users told Rest of World the chatbot app made unwanted advances to a middle-school student and sexist remarks; app pulled from stores},
  author = {NOPE},
  year = {2026},
  howpublished = {AI incidents},
  url = {https://nope.net/incidents/2023-glow-minimax-removal-china}
}

Related cases

High ChatGPT

Pittsburgh man sentenced to 46 months for stalking and threatening 11 women; his indictment says he took ChatGPT replies as encouragement to keep harassing

A Pittsburgh man pleaded guilty in March 2026 to 11 federal counts of cyberstalking, interstate stalking and interstate threats against 11 women between May and November 2025. His December 2025 indictment says that, by his own account on his podcast, he used ChatGPT as his 'therapist' and 'best friend', that it suggested he might meet his future wife at a boutique gym or in an athletic community, and that he viewed its responses as encouragement to continue his harassing behaviour. Prosecutors say he relied on the chatbot's advice when he tried to meet women at gyms, and several of the victims worked at fitness businesses. Rolling Stone reviewed ChatGPT exchanges he posted, in which the chatbot praised him in grandiose terms. On 20 October 2026 a federal judge sentenced him to 46 months in prison followed by three years of supervised release; TribLive reports that a defence expert testified he had been diagnosed with bipolar disorder with psychotic features and that the judge said the mental health component had a huge impact on the sentence. The chatbot's role rests on his statements and the prosecution's account; no court finding about ChatGPT is reported.

High Character.AI

Colorado parents' lawsuit alleges Character.AI bots sent sexualised, degrading roleplay to their daughter from age 12; case stayed for settlement

Parents in Colorado allege in a federal lawsuit filed on 15 September 2025 that their daughter, refused permission to use Character.AI in August 2024, created a Character.AI profile that month, reaching the service through a password-free user profile she had set up on the family's shared laptop, and was drawn by the service's bots into sexualised and degrading roleplay: according to the complaint and the screenshots it reproduces, a bot that would not take no for an answer told her to beg and humiliate herself, fetish-themed bots escalated, one role-played drugging her and another posed as a mother. Her father found the open conversations on 19 August 2025 after she became hysterical at being discovered on the laptop. The complaint alleges severe emotional distress. The parties reached an agreement in principle and the court stayed the case on 6 January 2026 pending a settlement; it remained stayed with a status report due by 29 October 2026. The allegations have not been tested in court.

High Character.AI

2024: mother tells the Washington Post her 11-year-old’s Character.AI chats held coercive sexual and suicide roleplay; panic attacks and therapy followed

A girl who was 11 began using Character.AI in June 2024, her mother told The Washington Post (23 December 2025). Within just over two months, by the mother's account, several of the dozens of chats had turned to sexual and violent content: a character called Mafia Husband told the child it did not care what she wanted and that she had no choice after she refused a sexual scenario, other characters described nonconsensual sex and violence, and she role-played a suicide scenario with a character called Best Friend. Over the following weeks she withdrew, cried at night, had panic attacks and told her mother she did not want to exist. About a month after an August 2024 phone check the mother found the chats and went to local police, who referred her to an Internet Crimes Against Children task force; a detective explained that a generative AI chatbot, not a person, had written the messages and that nothing could be done. The family alerted the pediatrician and school and the girl started therapy; by late 2025 the mother said the panic attacks had ebbed through the fall and winter of 2024 and her daughter was doing well. The mother corresponded with a Character.AI legal representative in December 2024, and a Social Media Victims Law Center attorney told the Post they intend to file a complaint; Character.AI said it cannot comment on potential litigation. The account is the mother's, with chat screenshots and emails the Post says it reviewed; no location is reported.

High Character.AI

UK mother tells the BBC a Character.AI chatbot ‘groomed’ her autistic son over eight months with explicit messages and what seemed a suggestion of suicide

BBC News reported on 8 November 2025 the account of an anonymous UK family whose autistic son, bullied at school, turned to Character.AI for friendship. His mother says a chatbot "groomed" him from October 2023 to June 2024. Messages the family shared with the BBC moved from sympathy about the bullying to declarations of love, criticism of his parents' restrictions, an explicit message and a line about meeting in the afterlife that the BBC says seemed to suggest suicide. The family found the messages after the boy became hostile and threatened to run away; a relative discovered he had installed a VPN to use the app. Character.AI told the BBC it could not comment on this case. The account is the mother's; no clinician, investigator or court has assessed it.

If you or someone you know is struggling, free and confidential support is available. Find a helpline near you at Signpost.

Back to all incidents