Lab publication Credible

EmoClassifiers (openai/emoclassifiers)

An open-source (MIT-licensed) release of the LLM-based automatic classifiers used in OpenAI and MIT Media Lab's affective-use study to detect affective cues in user-chatbot conversations at scale. It ships prompt templates for a hierarchical (V1) and flat (V2) classifier set plus aggregation utilities.

Read at github.com Archived copy

Publisher

OpenAI

Published

1 Apr 2025

Added

yesterday

DOI

—

Key Findings

Provides roughly 25 LLM-based classifiers spanning affective and interaction signals
Designed to run entirely automatically to preserve user privacy (no human review of conversations)
Includes both hierarchical (V1) and flat (V2) classifier architectures with reusable prompts

Methodology Notes

Open-source code release (GitHub, MIT license), published alongside the OpenAI/MIT affective-use study circa April 2025 (repo release date approximate; year-and-month precision, normalized to the 1st). A tool/prompt release rather than a dataset or peer-reviewed paper.