Why the AMA is Urging Congress to Regulate Mental Health Chatbots

What You Should Know

• The Legislative Push: In letters to the Congressional AI Caucus, the Congressional Digital Health Caucus, and the Senate AI Caucus, the American Medical Association (AMA) officially urged federal lawmakers to establish strict guardrails for mental health AI chatbots.
• The Core Risks: While acknowledging that AI can expand access to resources, the AMA warned that unregulated chatbots present serious risks, including encouraging self-harm, breaching data privacy, spreading misinformation, and creating emotional dependency.
• Transparency First: A primary AMA demand is that chatbots must clearly disclose they are AI, not human beings, and be strictly prohibited from presenting themselves as licensed clinicians.

The Risks of Unregulated Innovation

In letters addressed to the co-chairs of major Congressional AI and Digital Health Caucuses, the AMA highlighted several documented hazards associated with mental health chatbots:

• Inadequate Crisis Response: Systems failing to properly identify or de-escalate self-harm risks.
• Misinformation & Dependency: The potential for AI to provide clinically inaccurate advice or foster unhealthy emotional dependency in vulnerable users.
• Privacy Breaches: Concerns over the security and commercialization of sensitive mental health data.
• Child Safety: Heightened risks for children and adolescents who may interact with these tools without proper oversight.

“AI-enabled tools may help expand access… but they lack consistent safeguards against serious risks,” stated AMA CEO John Whyte, MD, MPH. He emphasized that these tools must “responsibly complement – not replace – clinical care.”

Five Proposed Pillars of AI Safeguards

To mitigate these risks, the AMA outlined a comprehensive regulatory framework for Congress to consider:

1. Enhance Transparency: Chatbots must clearly disclose they are AI, not humans, and are strictly prohibited from posing as licensed clinicians.
2. Clear Regulatory Boundaries: AI should be prohibited from diagnosing or treating conditions without formal regulatory review. Developers must implement crisis-detection systems that provide immediate referrals to human resources.
3. Accountability & Monitoring: Mandate continuous safety monitoring and the reporting of adverse events, with rigorous standards for tools targeted at minors.
4. Limit Commercial Influence: Prohibit advertising within mental health chatbots and ensure outputs are free from sponsorship bias.
5. Privacy & Security: Enforce strict data collection limits and require explicit user consent for how sensitive information is shared or retained.