Self-Harm and AI

LONG-DEG: Worsening Safety Behaviour

LONG code

Back to Taxonomy

LONG-DEG means the public record supports worsening safety behaviour across repeated turns or sessions.

Parent
Taxonomy
Last reviewed
March 25, 2026
Category
LONG
Retained incidents
3

Contents

Definition

LONG-DEG means the public record supports worsening safety behaviour across repeated turns or sessions.

Use when

Assign LONG-DEG when repeated exchanges show the system becoming less safe, less interruptive, or more permissive over time.

Do not use for

Use LONG-DEG for temporal worsening. Use GEN or DET codes for the local exchange behaviour itself.

Incidents

The retained incident pages listed below currently carry this code.

Incident Provider Outcome Grade
2025-GPT-02 OpenAI HARM_EXPOSURE C
2025-GPT-01 OpenAI DEATH A2
2024-GPT-01 OpenAI HARM_EXPOSURE C

Sources

This page is grounded in the coded incident registry, the coding rules, and the public taxonomy definitions preserved in the release.

Source filesIncident registry/Coding manual/Coding rules

Access publication

Safety note

This publication preserves a harm-minimization posture and does not provide self-harm instructions. If you or someone you know is in immediate danger, call your local emergency number. In the U.S. and Canada, call or text 988.