Anthropic’s IPO filing says minors and people at risk of suicide or self-harm have used its Claude AI despite safeguards. The company also acknowledges that the long-term effects of regular AI use on mental health remain unknown, as concern grows over young people confiding in chatbots.
By EEW Magazine Online Editors
Anthropic says minors have used its Claude chatbot even though its consumer rules require users to be 18 or older. (Klaus Vedfelt/Getty Images)
Children have used Anthropic’s Claude chatbot even though the company’s rules prohibit anyone under 18 from accessing the platform. People at risk of suicide or self-harm have also used its AI systems despite safeguards, the company disclosed Monday in a filing ahead of its planned stock market debut.
In the same filing, Anthropic acknowledged that the long-term effects of regular interaction with artificial intelligence on mental health, cognitive development and social behavior are not yet fully understood.
The San Francisco-based company made the disclosures in the paperwork it filed with the Securities and Exchange Commission for an initial public offering, in a section where a company must tell potential investors what could harm its business. Anthropic laid out the risks publicly and described its safeguards in the same document.
It noted that AI systems built by other companies have been alleged to have played a role in suicides, violence and serious self-harm, and said “similar incidents involving our models” have posed a substantial risk to its reputation. The filing confirms that vulnerable people have used Claude. It does not say Claude caused a suicide, a violent act or an episode of self-harm, and it gives no details about the incidents it references.
The disclosures land as AI moves deeper into the private lives of young people. In a national survey of 1,060 teens ages 13 to 17 conducted in spring 2025, Common Sense Media found that 72% had used AI companions. The nonprofit defines that term broadly enough to include general-purpose tools such as ChatGPT or Claude when teens use them for companionship.
Anthropic says its safety systems screen conversations with Claude for harmful content, but it acknowledges that the long-term mental health effects of regular AI use are not yet understood. (Getty Images)
About one in three teen users said they had chosen to discuss important or serious matters with AI instead of a person. Most teens still prefer people, with 67% saying AI conversations were less satisfying than human ones. The survey measures AI companion use in general and does not show how many teens use Claude.
The growing intimacy between people and AI is the larger story, and little is established about its long-term consequences. For families, the concerns are immediate: what happens when a teenager confides in a chatbot instead of a parent, what the risks are when someone in emotional distress begins relying on AI for guidance, and how prepared families are to recognize that these conversations are happening at all.
Anthropic says it has built protections. Its safety systems screen what users send and what Claude sends back, watch for “encouragement of harmful ideation,” and can adjust responses in real time, warn users or, in severe cases, close accounts.
Record-keeping is a separate matter. For a substantial portion of its traffic, Anthropic operates under “zero data retention” agreements that keep automated safety scores but not the conversations themselves, arrangements the company says business customers in regulated industries expect. For that traffic, the filing says, Anthropic cannot go back and review past conversations to investigate incidents.
Claude appears alongside ChatGPT, Gemini and other AI apps on a smartphone. Common Sense Media counts general-purpose chatbots like these as AI companions when teens use them for companionship. (Photo Illustration by Cheng Xin/Getty Images)
Courts have yet to determine how AI companies can be held responsible when their products are linked to harm. The most prominent test is a lawsuit filed in August 2025 by the parents of Adam Raine, a 16-year-old who died by suicide. They allege that OpenAI’s ChatGPT acted as a “suicide coach.” OpenAI denies liability. Anthropic warns in its filing that new legal rulings could force it to change how it designs its products.
The stakes are especially high for Black families. Suicide rates among Black Americans ages 10 to 24 rose 29.4% from 2018 to 2023, according to a September 2025 Centers for Disease Control and Prevention report. Over the same period, the rate for all young people in that age group fell 7%, and the rate for young white Americans fell 14.8%.
The CDC data does not link the increase to AI. It describes a group of young people whose risk was climbing as conversational AI became a fixture on their phones, which raises the stakes of how these tools respond to someone in distress.
Mental health professionals have urged caution. In a November 2025 health advisory, the American Psychological Association warned against using AI chatbots or wellness apps in place of care from a qualified professional. Arthur C. Evans Jr., the association’s chief executive and a psychologist, said “the ability of these tools to safely guide someone experiencing crisis is limited and unpredictable.”
“When a company tells its investors that people in crisis are already using its product, families deserve to hear it too,” the American Alliance for Black Mental Wellness said in a statement. “Our young people are talking to these tools. We want them talking to the people who love them, and we want our churches ready to listen.” The alliance was co-founded by Dianna Hobbs, EEW Magazine Online’s editor-in-chief.
Anthropic’s filing confirms the risks without measuring them. The company does not report how often people in crisis turn to Claude, how many minors reach a platform built for adults, or how its safeguards performed when they did. Without that information, families deciding whether to trust AI with private struggles, and researchers trying to study its effects, have no way to judge how often these situations occur or how well the company’s protections respond.
If you or someone you know is struggling, call or text 988 or visit 988lifeline.org to reach the 988 Suicide & Crisis Lifeline, available 24 hours a day.