All models do this last time I checked. Not just Chinese.

quantadev · 2025-01-27T19:22:37 1738005757

All closed-source models censor to the liking of their investors. Open Source models are generally less censored, but yeah DeepSeek is censored for sure.

hdjjhhvvhga · 2025-01-27T19:27:43 1738006063

Yes, but one would expect the Chinese not to fine-tune according to Western standards like in the quoted prompt.

girvo · 2025-01-28T01:41:58 1738028518

It becomes clearer when one realises that it was RLHF'd using ChatGPT and Qwen (and others I assume) output. It's caused it to "learn" the same weird Western censoring, and adopt the China-styled censoring too.

w4yai · 2025-01-27T18:47:06 1738003626

Wrong, at least with Claude : https://i.imgur.com/6dj1XAU.png

Palmik · 2025-01-27T18:57:32 1738004252

It's actually not wrong, here is the example from the OP: https://imgur.com/a/5uMYI44

girvo · 2025-01-28T01:40:46 1738028446

What government demanded them to do that under pain of being disappeared?

rsanek · 2025-01-28T00:50:37 1738025437

You get the same output from Claude regardless of gender. I wonder which government asked Anthropic to censor the output in this way

themgt · 2025-01-27T19:29:46 1738006186

"That's nothing. I can stand in front of the Kremlin and yell, 'To hell with Ronald Reagan,' too."