GPT-4chan 'hate speech machine'

GPT4-chan, an AI system trained using 3.3 million threads from 4chan's notorious 'politically incorrect' /pol/ board, has been posting thousands of sexist, racist, and anti-semitic slurs.

Developed by AI researcher Yannic Kilcher, GPT4-chan was released back onto 4chan as multiple bots, which posted freely on the board over several days over 30,000 times.

Critical response

The high volume of messages roused the suspicion of some 4chan users, who began to speculate on the identity of the anonymous poster.

Kilcher described GPT4-chan as a 'prank' to The Verge, arguing it 'perfectly encapsulated the mix of offensiveness, nihilism, trolling, and deep distrust of any information whatsoever that permeates most posts on /pol.'

AI researchers lambasted Kilcher’s project as nakedly self-promotional, naive, and unethical, calling out the fact that he tried to make the model freely accessible on AI community HuggingFace.

Transparency

They also pointed out that at no point were 4chan users informed they were interacting with an AI.

HuggingFace promptly restricted access to GPT4-chan.