FAIRT2V: Training-Free Debiasing for Text-to-Video Diffusion Models

TL;DR

FairT2V is a training-free method that reduces gender bias in text-to-video diffusion models by neutralizing prompt embeddings, improving fairness without compromising video quality.

Contribution

This work introduces a novel, training-free debiasing framework that targets encoder-induced bias in T2V models through prompt embedding neutralization and a dynamic debiasing schedule.

Findings

Significant reduction in gender bias across occupations.

Minimal impact on video quality.

Effective bias mitigation during early generation steps.

Abstract

Text-to-video (T2V) diffusion models have achieved rapid progress, yet their demographic biases, particularly gender bias, remain largely unexplored. We present FairT2V, a training-free debiasing framework for text-to-video generation that mitigates encoder-induced bias without finetuning. We first analyze demographic bias in T2V models and show that it primarily originates from pretrained text encoders, which encode implicit gender associations even for neutral prompts. We quantify this effect with a gender-leaning score that correlates with bias in generated videos. Based on this insight, FairT2V mitigates demographic bias by neutralizing prompt embeddings via anchor-based spherical geodesic transformations while preserving semantics. To maintain temporal coherence, we apply debiasing only during early identity-forming steps through a dynamic denoising schedule. We further propose a video-level fairness evaluation protocol combining VideoLLM-based reasoning with human verification. Experiments on the modern T2V model Open-Sora show that FairT2V substantially reduces demographic bias across occupations with minimal impact on video quality.