shipfeedAI news, curated daily

04:05:15 CET
28 JUL04:05:15shipfeed⋯
pull to refreshlast sync
Just in — 30 new
§ safety · storyline

OpenAI details GPT-Red, an AI system designed to find vulnerabilities in its own models

OpenAI details GPT-Red, an AI system designed to find vulnerabilities in its own models.

Jul 16 · · primary fetch2 sourcesupdated Jul 16 ·

OpenAI details GPT-Red, an AI system designed to find vulnerabilities in its own models SC Media

read full article on SC Media ↗
§ sources2 publications · timeline below
  1. SC MediaOpenAI details GPT-Red, an AI system designed to find vulnerabilities in its own modelsprimary
  2. MarkTechPostOpenAI Details GPT-Red: An Internal Automated Red-Teaming Model That Beat Human Red-Teamers 84% To 13% On Prompt Injection

§ how this story moved

  1. primary — MarkTechPost publishes the launch post.
  2. SC Media picks up coverage.