OpenAI ↗ · 2025-08-07 08:00 · UTC+8
From hard refusals to safe-completions: toward output-centric safety training
公司动态 OpenAI 原文语言:英文
Discover how OpenAI's new safe-completions approach in GPT-5 improves both safety and helpfulness in AI responses—moving beyond hard refusals to nuanced, output-centric safety training for handling dual-use prompts.
原文摘录
Discover how OpenAI's new safe-completions approach in GPT-5 improves both safety and helpfulness in AI responses—moving beyond hard refusals to nuanced, output-centric safety training for handling dual-use prompts.
以下是来源提供的简短摘录,完整内容请查看原文。