Introducing gpt-oss-safeguard

OpenAI Blog · 8 months ago

OpenAI released gpt-oss-safeguard, an open-weight model designed to classify content safety issues and let developers implement custom safety policies. The model is available as open weights, enabling developers to run it locally and adapt it to their specific safety requirements. Developers can now build customized content moderation systems without relying on closed APIs or OpenAI's default policies.