Achieved F1 score of 83%, enhancing safety of generative models.
Foundations for Safe Generative Media
Original: Runway Research | Foundations for Safe Generative Media
Importance: 生成AIモデルの安全性と公平性向上は、多くのユーザーに直接影響を与えるため。
Summary
Runway has announced guardrails for safety, fairness, and integrity in generative models to promote human creativity and support the media and entertainment industries. They have implemented a visual moderation system to detect malicious actors, achieving over 80% F1 score. They also have policies to protect children and aim to provide equitable experiences for all users.
Key Points
- Announced safety guardrails for generative models
- Achieved F1 score of 83% and recall of 88%
- Implemented policies to protect children
- Aiming for equitable experiences for all users
- Working on multilingual support
View developer notes (APIs, breaking changes, migration)
Runway has built an in-house visual moderation system that automatically detects and blocks inappropriate content, achieving F1 score of 83% and recall of 88%. This outperforms the best third-party API tested. They have also deployed solutions to reduce gender and racial biases in job prompts and are working towards multilingual support for their generative tools.
Source: https://runway.com/research/foundations-for-safe-generative-media
Outlet: Runway
This article is an AI-generated summary (OpenAI GPT-4o-mini) of publicly available information from Anthropic, OpenAI, Google, Meta, Mistral, DeepSeek, Sakana, and other vendors. The original source URL is always provided in accordance with fair-use citation requirements. Summaries are AI-generated and may contain mistranslations or misinterpretations. Always verify details with the original source.