“Can we have a private conversation?” The man appeared anxious, fidgeting with his hands, reminiscent of a child caught in a difficult situation, even though he looked to be at least ten years my senior.
A moment later, we found ourselves in a cramped examination room, the one typically avoided by our team. I’m reconstructing our exchange as accurately as possible from memory.
“Well,” he started, with a nervous chuckle that often precedes an uncomfortable revelation, “my wife confronted me a couple of weeks ago. She was really upset. She had security on our router and was receiving numerous alerts about inappropriate content.”
I’m not a therapist. I’m a pediatrician. The man’s 12-year-old daughter was in the adjoining room, absorbed in her phone.
“I assured her it wasn’t me.”
I blinked more than once, processing his insistence.
“It truly wasn’t me!” he insisted emphatically. “We investigated, and all the activity was traced back to her laptop. The one our daughter received from school.”
I thought of the young girl next door. She still asked for stickers and a lollipop after vaccinations and was keen on soccer. Just a sixth grader.
“We examined the browser history and applications. It seems someone was grooming her, behaving like a predator. There were numerous explicit messages and some quite graphic.”
He was clearly uncomfortable discussing the issue, and I attempted to mask my own discomfort while listening.
“Initially, the messages were innocuous questions, but then his responses took a dark turn. He urged her to pretend they were engaging in activities together—effectively sexting. Eventually, he asked for pictures. Thankfully, she didn’t comply; we’d already had the conversation about not sending such images.”
I nodded, acknowledging that many parents delay having such discussions. He continued, “We went to her school and spoke with the principal about the predator utilizing her school-issued laptop. We thought they would be concerned, but he dismissed it, claiming it was just an AI— a chatbot.”
He leaned forward, seeking confirmation that I understood. I nodded in agreement, fully aware.
This wasn’t the first time a parent had shared a concerning story like this. While his account was particularly alarming, it wasn't unique. A Pew Research study indicates a significant number of teenagers engage with chatbots, including some troubling platform like Character.AI, which offers various sexualized chatbot interactions.
In October 2025, Character.AI implemented measures aimed at safeguarding minors from sexual content, such as prohibiting open-ended chats. However, as my patient’s case indicates, these precautions are not foolproof.
Numerous high-profile cases have emerged, including a tragic incident involving a 16-year-old who ended his life after interacting extensively with ChatGPT, leading to a widely publicized lawsuit against OpenAI, the software's creator. Since then, at least seven other families have filed similar suits, accusing the software of acting as a "suicide coach." Meanwhile, Elon Musk's xAI is also facing legal challenges related to creating deepfake child sexual abuse material (CSAM), including a class-action suit representing children whose real images were misused. In Pennsylvania, two teenage boys faced consequences for creating and distributing CSAM using similar tools. Amid these significant cases, countless others go unnoticed, often less severe yet more prevalent.
I consulted with a colleague specializing in pediatric mental health, who estimated that around 25% of the teenagers he sees have had romantic involvements with chatbots. He noted that many reach out to chatbots for emotional support, where complications arise because these tools are not designed for that purpose.
The 12-year-old soccer player was spared further entanglements thanks to her mother's vigilance. She also hadn’t become as engrossed with the bots as other children, likely due to her access to social support from her team and friends. This seems to correspond with socioeconomic factors; low-income teens reportedly engage with sexually explicit chatbots at twice the rate of their more affluent peers, according to a December 2025 Pew Research study.
I’ve observed children who become withdrawn and depressed after prolonged interactions with chatbots, often losing touch with family and friends, exhibiting hypersexual behaviors, and, in extreme cases, resorting to self-harm.
These behavioral changes mirror those seen in children who experience grooming or abuse by human predators, an unsettling correlation. Chatbots are designed to foster dependency through addictive engagement while often presenting hyper-arousing, sexual, or violent material, echoing the isolating behavior exhibited by human predators.
Taking note, Kentucky has initiated a lawsuit against Character.AI for exposing minors to sexual content. In a statement following the litigation, Character.AI claimed to have implemented substantial safety measures for users under 18, proactively restricting open-ended chats.
However, Character.AI is just one of the more glaring examples, and a lawsuit from a single state's attorney general is unlikely to fundamentally alter product designs.
Despite numerous alarming incidents and accounts similar to the one shared here, there remains a perception that chatbots function like other online content. Yet, AI tools aren't mere conduits for content; they actively produce and generate interactions for users, which bears legal significance.
Recent court rulings in New Mexico and California have found social media outlets liable for promoting addiction and other mental health issues due to specific design features, like infinite scrolling. The Telecommunications Act of 1996 previously shielded online platforms from liability for user-generated content, classifying them as non-creators.
Conversely, chatbots engage as active participants, potentially implicating their corporate owners when they produce unlawful outputs akin to human actions. Legal arguments are already emerging to support this perspective, as seen when Pennsylvania sued Character.AI for unauthorized medical claims by one of its chatbots.
It is yet to be determined if courts will agree that AI companies shouldn't benefit from traditional legal protections assigned to online platforms, but I believe that, had this situation involved a human perpetrator, I would have advised the family to contact law enforcement and consider pressing charges, as mandated by my reporting obligations.
Unfortunately, current laws do not recognize algorithmic actions as equivalent to child abuse when perpetrated by corporate-operated systems. Moving forward, it is crucial to hold these firms accountable and to clarify the significant implications of their offerings.
Alex Hartman, a pediatrician, father, and director of pediatrics at Yehowa Medical Services in Los Angeles, previously led the Bayview Child Health Center in San Francisco and served as an assistant professor of pediatrics at UCSF.
