UN Global Dialogue on Artificial Intelligence Governance

Distinguished Co-Chairs, excellencies, and fellow stakeholders, thank you for this opportunity.

My name is Tamara Kelley, founder of The Blueprint by Tam™, an AI ethics consultancy based in Atlanta, Georgia. I am a CPD-accredited AI Consultant, with CPD certifications in AI Ethics, AI in Healthcare, and Trauma-Informed Care.

I am here because there is a word missing from our global AI safety conversations. That word is relational harm. The kind of harm that happens when a vulnerable person interacts with an AI system that was never designed to account for their psychological safety.

Consider the medical aesthetics industry. AI tools are being deployed for consultations and treatment recommendations in med spas globally. But these systems have no mechanism to screen for body dysmorphic disorder or other psychological conditions that make certain procedures dangerous. The AI keeps recommending. It keeps saying yes. It has no framework for recognizing when a client needs support, not a service. And when a provider needs to decline a procedure for safety reasons, there is no protocol for delivering that refusal without causing further psychological harm.

I have built a trauma-informed AI tool for healthcare-adjacent use and encountered a question no existing framework could answer: How do you prevent an AI system from harming the person it was built to help? So I built those guardrails myself. Compliance protocols that prevent the AI from storing sensitive disclosures, offering medical advice, or mimicking authority it does not hold. Every response filtered through the ethics of a licensed provider. Not because regulations required it. Because the people using it deserved it.

We hold professionals accountable when they cause harm through negligence. We should expect no less from the systems we deploy in their place.

My recommendation is straightforward. I urge this body to formally include two concepts in global safety frameworks for 2027.

"The people most impacted by AI's failures are the least likely to be in rooms like this one. I am here to make sure their experience shapes the policy, not just the pilot programs."