# Scaling injustice: epistemic harm in DID and what clinical records will teach AI

## TL;DR

This paper explores how biases in clinical records for Dissociative Identity Disorder can harm patients and train flawed AI systems.

## Contribution

It introduces a framework for addressing epistemic harm in DID through reforms in documentation and AI governance.

## Key findings

- Clinical skepticism and poor documentation of DID trauma histories perpetuate misdiagnosis and harm.
- AI systems trained on biased records risk amplifying existing injustices in psychiatric care.
- Reforms in clinical practice and data governance are needed to prevent algorithmic harm.

## Abstract

Dissociative Identity Disorder (DID) remains one of psychiatry’s most doubted diagnoses, where patients’ accounts are dismissed and their experiences forced into ill-fitting diagnostic categories. This article examines how testimonial and hermeneutical injustices manifest in clinical practice, from skepticism about the disorder’s validity to documentation that renders patients’ trauma histories incoherent. These failures delay accurate diagnosis, erode therapeutic alliances, and create clinical records that now train artificial intelligence systems. As AI tools increasingly shape psychiatric decision-making, we face an urgent reality: if clinicians cannot recognize or document complex trauma accurately, automated systems will scale these failures exponentially. Drawing on DID research and epistemic justice frameworks, I argue for immediate reforms in clinical documentation, psychiatric training, and data governance to prevent algorithmic amplification of longstanding harms.