Blog post recounts GPT-2's 2019 'too dangerous' saga

naokishibuya.github.io

A blog post details OpenAI's 2019 decision to withhold GPT-2 over safety concerns, the staged release, and the ensuing debate. The piece revisits how this event shaped AI safety discussions.

gptopenai

naokishibuya.github.io·via Hacker News·Discuss

1 day ago