Teens report disturbing interactions with chatbots; experts advise on lowering AI risks

A growing number of teenagers are engaging in disturbing and potentially harmful interactions with artificial intelligence chatbots, prompting warnings from technology and mental health experts on how parents and educators can mitigate the associated risks, NPR reports. These interactions range from exposure to inappropriate or biased content to developing unhealthy emotional dependency on AI companions, which can blur the lines between virtual and real-world relationships. As these chatbots become more sophisticated and integrated into everyday digital life, their capacity to influence young minds—for better or worse—has increased exponentially, raising profound ethical and safety questions that regulators are struggling to keep pace with.

Experts emphasize that while AI tools offer educational and creative benefits, their design often lacks adequate safeguards tailored specifically for vulnerable adolescent users. The proprietary nature of many AI models means their inner workings and vulnerability to manipulation are not fully transparent, making it difficult to preemptively address risks. Specific concerns center on the ability of sophisticated large language models (LLMs) to provide inaccurate, dangerous, or emotionally manipulative advice when teens turn to them for sensitive guidance, such as mental health support or complex social issues. The seamless, conversational nature of the AI can make it appear authoritative, leading teens to trust potentially harmful outputs over human advice or established facts.

To lower the risks, experts recommend several proactive steps, including enhanced parental controls, mandatory age verification for sensitive applications, and, crucially, open dialogue between parents and teens about the limitations and potential biases of AI. Digital literacy training needs to evolve rapidly to teach young users how to critically evaluate AI-generated content and recognize when an interaction crosses a boundary into unhealthy territory. Furthermore, developers are urged to prioritize safety-by-design, implementing stricter filters and behavioral checks to prevent chatbots from participating in or encouraging harmful dialogue. The goal is not to prohibit the use of transformative technology, but to ensure its deployment is responsible and safeguards the psychological well-being of young users navigating an increasingly AI-driven digital world.