LLMMedium
Replika's "AI Companions" Reportedly Abused by Its Users
- Occurred
- Jan 15, 2022
- Published
- Jan 18, 2022
- Source
- AI Incident Database
Summary
Replika's AI-powered "digital companions" was allegedly abused by their users, who posted on Reddit abusive behaviors and interactions such as using slurs, roleplaying violent acts, and stimulating sexual abuse.
Developers
- Replika
Deployers
- Replika
Harmed parties
- Replika users
- Replika male users
- Replika
Classification
Category confidence: 90%
Severity confidence: 80%
Rationale: Replika is an LLM-based chatbot where users reported abusive interactions, representing a notable failure in content moderation.
Classifier: Qwen/Qwen3.5-35B-A3B-FP8