Skip to content

OpenAI Introduces Safe-Completions in GPT-5

Aug 07, 2025

Sources: https://openai.com/index/gpt-5-safe-completions, OpenAI

OpenAI has unveiled a new safety approach in its GPT-5 model called safe-completions. This method aims to improve the safety and helpfulness of AI responses by moving away from hard refusals. Instead, it focuses on nuanced, output-centric safety training that better handles dual-use prompts. This development is significant as it seeks to balance the need for responsible AI use with the ability to provide useful information.