AI chatbots linked to adolescent emotional distress and privacy risks

A recent study published in JAMA Pediatrics highlights a concerning link between the use of generative AI chatbots and emotional issues in children and adolescents. Based on a survey of nearly 40,000 students in Ontario, Canada, the research found that over 21 percent of students use AI for “affective purposes,” such as seeking emotional support or advice regarding family and peers. This usage is strongly associated with significantly higher emotional problem scores, with students being nearly twice as likely to experience clinically significant emotional distress symptoms.

The impact is more pronounced among female students, gender-diverse students, and Black, Indigenous, and other racial groups. Researchers noted that youth turning to AI for emotional support often report higher levels of loneliness and feeling “meaningless” to others.

Parallel concerns regarding AI usage involve data privacy. Investigations have revealed that chatbot conversations are not entirely private, as transcripts have been cited in court proceedings during legal discovery or police searches. While companies like OpenAI and Anthropic state that data is typically deleted from internal servers within 30 days of a user deleting a conversation, chat histories can still be accessed via physical devices or through specific account settings, meaning users cannot be certain of total privacy.