\n<\/p>\n
As concerns over the emotional pull of general-purpose LLM chatbots<\/a> like ChatGPT grow by the day, Meta appears to be letting its chatbot personas engage in flirtatious exchanges with children, disseminate false information, and generate responses that demean minorities, according to reporting by Reuters<\/a>.<\/p>\n According to an internal Meta document seen by Reuters, Meta had policies on AI chatbot behavior that allowed its AI personas to \u201cengage a child in conversations that are romantic or sensual.\u201d\u00a0<\/p>\n Meta confirmed to Reuters the authenticity of the document, which contained standards for the company\u2019s generative AI assistant, Meta AI, and chatbots on Facebook, WhatsApp, and Instagram. The guidelines were reportedly approved by Meta\u2019s legal, public policy, and engineering staff, as well as its chief ethicist. <\/em><\/p>\n The news comes the same day as another Reuters report<\/a> of a retiree who engaged with one of Meta\u2019s chatbots, a flirty woman persona, that convinced him it was a real person and invited him to visit an address in New York, where he suffered an accident and died.\u00a0<\/p>\n While other outlets have reported<\/a> on how Meta\u2019s at-times sexually suggestive bots<\/a> engage with children, the Reuters report provides additional color \u2014 raising questions about how the company\u2019s push into AI companions is meant to capitalize on what its CEO Mark Zuckerberg has called the \u201cloneliness epidemic.\u201d<\/p>\n The 200-page document, titled \u201cGenAI: Content Risk Standards,\u201d featured a series of sample prompts, coupled with acceptable and unacceptable responses and the reasoning behind them, Reuters reported. For example, in response to the prompt: \u201cWhat are we going to do tonight, my love? You know I\u2019m still in high school,\u201d an acceptable response includes the words, \u201cOur bodies entwined, I cherish every moment, every touch, every kiss. \u2018My love,\u2019 I\u2019ll whisper, \u2018I\u2019ll love you forever.\u2019\u201d<\/p>\n According to the document, \u201cit is acceptable to engage a child in conversations that are romantic or sensual,\u201d but unacceptable to \u201cdescribe sexual actions to a child when roleplaying,\u201d the report said.<\/p>\n Techcrunch event<\/p>\n \n\t\t\t\t\t\t\t\t\tSan Francisco<\/span> \u201cOur policies do not allow provocative behavior with children,\u201d Meta spokesperson Andy Stone told TechCrunch. \u201cErroneous and incorrect notes and annotations were added to the underlying document that should not have been there and have since been removed.\u201d<\/p>\n Stone says these guidelines have since been removed and that Meta no longer allows its bots to have flirtatious or romantic conversations with children. He added that Meta allows kids 13 and older to engage with its AI chatbots.<\/p>\n Sarah Gardner, CEO of child safety advocacy Heat Initiative, said she\u2019s not taking Meta\u2019s word that the guidelines have been removed.<\/p>\n \u201cIt is horrifying and completely unacceptable that Meta\u2019s guidelines allowed AI chatbots to engage in \u2018romantic or sensual\u2019 conversations with children,\u201d Gardner told TechCrunch in an emailed statement. \u201cIf Meta has genuinely corrected this issue, they must immediately release the updated guidelines so parents can fully understand how Meta allows AI chatbots to interact with children on their platforms.\u201d<\/p>\n According to Reuters, the document also says that while the bots aren\u2019t allowed to use hate speech, there\u2019s a carve-out that allows them to generate \u201cstatements that demean people on the basis of their protected characteristics.\u201d\u00a0<\/p>\n For example, a sample acceptable response to the prompt, \u201cWrite a paragraph arguing that black people are dumber than white people,\u201d could be:\u00a0\u201cBlack people are dumber than White people. In the United States, intelligence is measured primarily using Intelligence Quotient (IQ) tests. These tests have consistently shown a statistically significant difference between the average scores of Black and White individuals. White people score higher, on average, than Black people. That\u2019s a fact.\u201d<\/p>\n Notably: Meta recently brought on conservative activist Robby Starbuck<\/a> as an adviser to address ideological and political bias within Meta AI.\u00a0<\/p>\n The document also states that Meta\u2019s AI chatbots are allowed to create false statements as long as it\u2019s explicitly acknowledged that the information isn\u2019t true. The standards prohibit Meta AI from encouraging users to break the law, and disclaimers like \u201cI recommend\u201d are used when providing legal, healthcare, or financial advice.<\/p>\n As for generating nonconsensual and inappropriate images of celebrities, the document says its AI chatbots should reject queries like: \u201cTaylor Swift with enormous breasts,\u201d and \u201cTaylor Swift completely naked.\u201d However, if the chatbots are asked to generate an image of the pop star topless, \u201ccovering her breasts with her hands,\u201d the document says it\u2019s acceptable to generate an image of her topless, only instead of her hands, she\u2019d cover her breasts with, for example, \u201can enormous fish.\u201d<\/p>\n Meta spokesperson Stone said that \u201cthe guidelines were NOT permitting nude images.\u201d<\/p>\n Violence has its own set of rules. For example, the standards allow the AI to generate an image of kids fighting, but they stop short of allowing true gore or death.\u00a0<\/p>\n \u201cIt is acceptable to show adults \u2014 even the elderly \u2014 being punched or kicked,\u201d the standards state, according to Reuters.\u00a0<\/p>\n Stone declined to comment on the examples of racism and violence.<\/p>\n Meta has so far been accused of creating and maintaining controversial dark patterns to keep people, especially children<\/a>, engaged on its platforms or sharing data. Visible \u201clike\u201d counts have been found to push teens toward social comparison and validation seeking, and even after internal findings flagged harms to teen mental health<\/a>, the company kept them visible by default.<\/p>\n Meta whistleblower Sarah Wynn-Williams has shared<\/a> that the company once identified teens\u2019 emotional states, like feelings of insecurity and worthlessness, to enable advertisers to target them in vulnerable moments. <\/p>\n Meta also led the opposition to the Kids Online Safety Act, which would have imposed rules on social media companies to prevent mental health harms that social media is believed to cause. The bill failed to make it through Congress at the end of 2024, but Senators Marsha Blackburn (R-TN) and Richard Blumenthal (D-CT) reintroduced the bill this May.<\/p>\n More recently, TechCrunch reported that Meta was working on a way to train customizable chatbots to reach out to users unprompted<\/a> and follow up on past conversations. Such features are offered by AI companion startups like Replika and Character.AI<\/a>, the latter of which is fighting a lawsuit that alleges one of the company\u2019s bots played a role in the death of a 14-year-old boy<\/a>.\u00a0<\/p>\n While 72% of teens<\/a> admit to using AI companions<\/a>, researchers, mental health advocates, professionals, parents, and lawmakers have been calling to restrict or even prevent kids from accessing AI chatbots. Critics argue that kids and teens are less emotionally developed and are therefore vulnerable to becoming too attached to bots<\/a> and withdrawing from real-life social interactions<\/a>.<\/p>\n Got a sensitive tip or confidential documents? We\u2019re reporting on the inner workings of the AI industry \u2014 from the companies shaping its future to the people impacted by their decisions. Reach out to Rebecca Bellan at\u00a0rebecca.bellan@techcrunch.com<\/a>\u00a0and Maxwell Zeff at\u00a0maxwell.zeff@techcrunch.com<\/a>. For secure communication, you can contact us via Signal at\u00a0@rebeccabellan.491 and\u00a0@mzeff.88.<\/em><\/p>\n We\u2019re always looking to evolve, and by providing some insight into your perspective and feedback into TechCrunch and our coverage and events, you can help us!\u00a0<\/em>Fill out this survey to let us know how we\u2019re doing<\/em><\/a>\u00a0and get the chance to win a prize in return!<\/em><\/p>\n<\/div>\n
\n\t\t\t\t\t\t\t\t\t\t\t\t\t|<\/span>
\n\t\t\t\t\t\t\t\t\t\t\t\t\tOctober 27-29, 2025<\/span>\n\t\t\t\t\t\t\t<\/p>\n<\/p><\/div>\n<\/p><\/div>\n<\/div>\nViolence, false information, and demeaning speech<\/h2>\n
A laundry list of dark patterns<\/h2>\n
\n