
The Washington Post @washingtonpost
The work is controversial. Some language models have shown themselves to be racist or sexist, or easily manipulated into spouting hate speech or lies when prompted with the right statements or questions. https://t.co/YfWCWio0wm — PolitiTweet.org