GPT-2: Too Dangerous To Release — A 2019 Retrospective
Hacker News (AI keywords)·5 days ago·Commentary
In 2019, OpenAI staged the release of GPT-2, citing fears it could enable large-scale disinformation and spam generation. The move sparked debate: was it responsible AI safety practice or a savvy PR stunt? Written in late 2022, this blog post revisits the episode now that GPT-2 looks quaint compared to GPT-3/4, asking whether the original fears were justified.
