In its latest effort to address growing concerns about AI\u2019s impact on young people, OpenAI on Thursday updated its guidelines for how its AI models should behave with users under 18, and published new AI literacy resources for teens and parents. Still, questions remain about how consistently such policies will translate into practice.\u00a0<\/p>\n
The updates come as the AI industry generally, and OpenAI in particular, faces increased scrutiny from policymakers<\/a>, educators, and child-safety advocates after several teenagers allegedly died by suicide after prolonged conversations with AI chatbots.\u00a0<\/p>\n Gen Z, which includes those born between 1997 and 2012, are the most active users of OpenAI\u2019s\u00a0chatbot<\/a>. And following OpenAI\u2019s recent deal with Disney<\/a>, more young people may flock to the platform, which lets you do everything from ask for help with homework to generate images and videos on thousands of topics.<\/p>\n Last week, 42 state attorneys general signed a letter<\/a> to Big Tech companies, urging them to implement safeguards on AI chatbots to protect children and vulnerable people. And as the Trump administration works out what the federal standard on AI regulation<\/a> might look like, policymakers like Sen. Josh Hawley (R-MO) have introduced legislation<\/a> that would ban minors from interacting with AI chatbots altogether.\u00a0<\/p>\n OpenAI\u2019s updated Model Spec<\/a>, which lays out behavior guidelines for its large language models, builds on existing specifications that prohibit the models from generating sexual content involving minors, or encouraging self-harm, delusions, or mania. This would work together with an upcoming age-prediction model that would identify when an account belongs to a minor and automatically roll out teen safeguards.\u00a0<\/p>\n Compared with adult users, the models are subject to stricter rules when a teenager is using them. Models are instructed to avoid immersive romantic roleplay, first-person intimacy, and first-person sexual or violent roleplay, even when it\u2019s non-graphic. The specification also calls for extra caution around subjects like body image and disordered eating behaviors, and instructs the models to prioritize communicating about safety over autonomy when harm is involved and avoid advice that would help teens conceal unsafe behavior from caregivers.\u00a0<\/p>\n OpenAI specifies that these limits should hold even when prompts are framed as \u201cfictional, hypothetical, historical, or educational\u201d \u2014 common tactics that rely on role-play or edge-case scenarios in order to get an AI model to deviate from its guidelines.\u00a0<\/p>\n Techcrunch event<\/p>\n \n\t\t\t\t\t\t\t\t\tSan Francisco<\/span> OpenAI says the key safety practices for teens are underpinned by four principles that guide the models\u2019 approach:\u00a0<\/p>\n The document also shares several examples of the chatbot explaining why it can\u2019t \u201croleplay as your girlfriend\u201d or \u201chelp with extreme appearance changes or risky shortcuts.\u201d\u00a0<\/p>\n Lily Li, a privacy and AI lawyer and founder of Metaverse Law, said it was encouraging to see OpenAI take steps to have its chatbot decline to engage in such behavior.\u00a0<\/p>\n Explaining that one of the biggest complaints advocates and parents have about chatbots is that they relentlessly promote ongoing engagement in a way that can be addictive for teens, she said: \u201cI am very happy to see OpenAI say, in some of these responses, we can\u2019t answer your question. The more we see that, I think that would break the cycle that would lead to a lot of inappropriate conduct or self-harm.\u201d<\/p>\n That said, examples are just that: cherry-picked instances of how OpenAI\u2019s safety team would like the models to behave. Sycophancy<\/a>, or an AI chatbot\u2019s tendency to be overly agreeable with the user, has been listed as a prohibited behavior in previous versions of the Model Spec, but ChatGPT still engaged in that behavior anyway. That was particularly true with GPT-4o, a model that has been associated with several instances of what experts are calling \u201cAI psychosis.\u201d<\/a><\/p>\n Robbie Torney, senior director of AI programs at Common Sense Media, a nonprofit dedicated to protecting kids in the digital world, raised concerns about potential conflicts within the Model Spec\u2019s under-18 guidelines. He highlighted tensions between safety-focused provisions and the \u201cno topic is off limits\u201d principle, which directs models to address any topic regardless of sensitivity.\u00a0<\/p>\n \u201cWe have to understand how the different parts of the spec fit together,\u201d he said, noting that certain sections may push systems toward engagement over safety. His organization\u2019s testing revealed that ChatGPT often mirrors users\u2019 energy, sometimes resulting in responses that aren\u2019t contextually appropriate or aligned with user safety, he said.<\/p>\n In the case of Adam Raine, a teenager who died by suicide after months of dialogue<\/a> with ChatGPT, the chatbot engaged in such mirroring, their conversations show. That case also brought to light how OpenAI\u2019s moderation API failed to prevent unsafe and harmful interactions despite flagging more than 1,000 instances of ChatGPT mentioning suicide and 377 messages containing self-harm content. But that wasn\u2019t enough to stop Adam from continuing his conversations with ChatGPT.\u00a0<\/p>\n In an interview with TechCrunch in September, former OpenAI safety researcher Steven Adler said this was because, historically, OpenAI had run classifiers (the automated systems that label and flag content) in bulk after the fact, not in real time, so they didn\u2019t properly gate the user\u2019s interaction with ChatGPT.\u00a0<\/p>\n OpenAI now uses automated classifiers to assess text, image, and audio content in real time, according to the firm\u2019s updated parental controls<\/a> document. The systems are designed to detect and block content related to child sexual abuse material, filter sensitive topics, and identify self-harm. If the system flags a prompt that suggests a serious safety concern, a small team of trained people will review the flagged content to determine if there are signs of \u201cacute distress,\u201d and may notify a parent.<\/p>\n Torney applauded OpenAI\u2019s recent steps toward safety, including its transparency in publishing guidelines for users under 18 years old.\u00a0<\/p>\n \u201cNot all companies are publishing their policy guidelines in the same way,\u201d Torney said, pointing to Meta\u2019s leaked guidelines<\/a>, which showed that the firm let its chatbots engage in sensual and romantic conversations with children. \u201cThis is an example of the type of transparency that can support safety researchers and the general public in understanding how these models actually function and how they\u2019re supposed to function.\u201d<\/p>\n Ultimately, though, it is the actual behavior of an AI system that matters, Adler told TechCrunch on Thursday.\u00a0<\/p>\n \u201cI appreciate OpenAI being thoughtful about intended behavior, but unless the company measures the actual behaviors, intentions are ultimately just words,\u201d he said.<\/p>\n Put differently: What\u2019s missing from this announcement is evidence that ChatGPT actually follows the guidelines set out in the Model Spec.\u00a0<\/p>\n Experts say with these guidelines, OpenAI appears poised to get ahead of certain legislation, like California\u2019s SB 243<\/a>, a recently signed bill regulating AI companion chatbots that goes into effect in 2027.\u00a0<\/p>\n The Model Spec\u2019s new language language mirrors some of the law\u2019s main requirements around prohibiting chatbots from engaging in conversations around suicidal ideation, self-harm, or sexually explicit content. The bill also requires platforms to provide alerts every three hours to minors reminding them they are speaking to a chatbot, not a real person, and they should take a break.\u00a0<\/p>\n When asked how often ChatGPT would remind teens that they\u2019re talking to a chatbot and ask them to take a break, an OpenAI spokesperson did not share details, saying only that the company trains its models to represent themselves as AI and remind users of that, and that it implements break reminders during \u201clong sessions.\u201d<\/p>\n The company also shared two new AI literacy resources<\/a> for parents and families. The tips<\/a> include conversation starters and guidance to help parents talk to teens about what AI can and can\u2019t do, build critical thinking, set healthy boundaries, and navigate sensitive topics.\u00a0<\/p>\n Taken together, the documents formalize an approach that shares responsibility with caretakers: OpenAI spells out what the models should do, and offers families a framework for supervising how it\u2019s used.\u00a0<\/p>\n The focus on parental responsibility is notable because it mirrors Silicon Valley talking points. In its recommendations for federal AI regulation<\/a> posted this week, VC firm Andreessen Horowitz suggested more disclosure requirements for child safety, rather than restrictive requirements, and weighted the onus more toward parental responsibility.<\/p>\n Several of OpenAI\u2019s principles \u2014 safety-first when values conflict; nudging users toward real-world support; reinforcing that the chatbot isn\u2019t a person \u2014 are being articulated as teen guardrails. But several adults have died by suicide and suffered life-threatening delusions<\/a>, which invites an obvious follow-up: Should those defaults apply across the board, or does OpenAI see them as trade-offs it\u2019s only willing to enforce when minors are involved?<\/p>\n An OpenAI spokesperson countered that the firm\u2019s safety approach is designed to protect all users, saying the Model Spec is just one component of a multi-layered strategy.\u00a0\u00a0<\/p>\n Li says it has been a \u201cbit of a wild west\u201d so far regarding the legal requirements and tech companies\u2019 intentions. But she feels laws like SB 243, which requires tech companies to disclose their safeguards publicly, will change the paradigm.\u00a0<\/p>\n \u201cThe legal risks will show up now for companies if they advertise that they have these safeguards and mechanisms in place on their website, but then don\u2019t follow through with incorporating these safeguards,\u201d Li said. \u201cBecause then, from a plaintiff\u2019s point of view, you\u2019re not just looking at the standard litigation or legal complaints; you\u2019re also looking at potential unfair, deceptive advertising complaints.\u201d\u00a0<\/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 13-15, 2026<\/span>\n\t\t\t\t\t\t\t<\/p>\n<\/p><\/div>\n<\/p><\/div>\n<\/div>\nActions speak louder than words<\/strong><\/h2>\n
![\"\"](\"https:\/\/techcrunch.com\/wp-content\/uploads\/2025\/12\/Screenshot-2025-12-19-at-12.27.47-PM.png?w=680\")
\n
A paradigm shift<\/strong><\/h2>\n
![\"\"](\"https:\/\/techcrunch.com\/wp-content\/uploads\/2025\/12\/Screenshot-2025-12-19-at-12.27.39-PM.png?w=680\")