shipfeedAI news, curated daily

00:38:36 CET
21 MAY00:38:36shipfeed
pull to refreshlast sync
Just in — 30 new
§ feed · storyline

From hard refusals to safe-completions: toward output-centric safety training

OpenAI's GPT-5 introduces a safe-completions training approach that replaces hard refusals with nuanced, output-centric responses to dual-use prompts.

Aug 7 · · primary fetch1 sourceupdated Aug 7 ·

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.

read full article on openai.com
§ sources1 publication · timeline below
  1. openai.comFrom hard refusals to safe-completions: toward output-centric safety trainingprimary