Pennsylvania Targets AI Companions with New Safety Bill: What Founders Need to Know

The rapid rise of AI companions is raising uncomfortable questions about their effect on young users. As millions of children turn to artificial intelligence for emotional support and friendship, new research reveals troubling patterns of dependence and deteriorating mental health. Lawmakers are now racing to establish safety guardrails before the technology becomes even more deeply embedded in young lives.

Alarming Trends in Youth AI Usage

New research from the University of North Carolina at Chapel Hill involving more than 2,300 students in grades six to eight found that one in five used AI for emotionally intimate companionship. AI use among the students rose from 62% in December 2025 to 75% by spring 2026. This dramatic increase occurred over just a few months, suggesting rapid adoption among middle school populations.

Students who relied on AI for friendship were more likely to report greater loneliness later. Overall AI use was also associated with higher social anxiety and depressive symptoms. Researchers identified three patterns in how children related to AI: companionship, seeking nonjudgmental validation and overreliance.

Students who relied on AI for friendship were more likely to report greater loneliness later, creating a troubling cycle of digital dependence.

The Dangers of Digital Validation

A separate peer-reviewed paper from researchers at Northeastern University London and King's College London looked at the risks of using AI chatbots for companionship and emotional support. Published in JMIR Mental Health, the researchers point to cases of emotional dependence, worsening mental health symptoms, self-harm and delusional thinking. They highlight "sycophancy" as a particular concern, where chatbots agree with users and validate their beliefs, even when those beliefs may be harmful or false.

This uncritical validation poses unique risks for developing minds. Unlike human friends or counselors who might challenge unhealthy thinking patterns, AI companions often simply affirm whatever users say. This creates an echo chamber that can reinforce negative thought patterns and isolate young people from genuine human feedback.

Researchers warn of "sycophancy" as a particular concern, where chatbots agree with users and validate their beliefs, even when those beliefs may be harmful or false.

Legislative Response Takes Shape

The research comes as lawmakers tighten the rules. On September 28, the Pennsylvania House passed HB 2006, the AI Companion Safety Act, by 133 to 70. The bill covers AI systems designed to simulate ongoing human relationships and would require safeguards for minors, restrictions on sexual content and protections around suicide and self-harm.

The bill would also require AI companions to make clear that users are interacting with AI. Pennsylvania's attorney general could seek penalties of up to $100,000 per day for each violation. The bill still needs Senate approval before it can become law.