Skip to content
BiasMedium

Facebook's Automated Tools Failed to Adequately Remove Hate Speech, Violence, and Incitement

Occurred
Mar 1, 2021
Published
Oct 18, 2021
Source
AI Incident Database

Summary

Facebook's automated moderation tools were shown by internal documents performing incomparably to human moderators, and accounting for only a small fraction of hate speech, violence, and incitement content removal.

Developers

  • Facebook

Deployers

  • Facebook

Harmed parties

  • Facebook users

Classification

Category confidence: 85%
Severity confidence: 90%

Rationale: Automated moderation tools failed to remove hate speech, indicating potential bias and disparate impact in content moderation.

Classifier: Qwen/Qwen3.5-35B-A3B-FP8

Visit source