Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Questions censored by DeepSeek

promptfoo.dev

Post reply on HN
Print view View original

11–20 of 257 posts

Re: Questions censored by DeepSeek

#11

Post by darth_avocado »

TOMDM wrote: post #3
tehjoker wrote: post #2

[flagged]

The What's Next section at the bottom seems to deliver a fairly balanced perspective. > What's Next DeepSeek-R1 is impressive, but its utility is clouded by concerns over censorship and the use of user data for training. The censorship is not unusual for Chinese models. It seems to be applied by brute force, which makes it easy to test and detect. It will matter less once models similar to R1 are reproduced without t…

> The censorship is not unusual for Chinese models.

It is not unusual for pretty much any model. It’s fair to say any model will be culturally representative of the people who built it. There have been criticisms around models built in US censoring certain things based on politics that are US centric that I am sure the Chinese model will not be censoring. And I am also certain that the censorship may also have overlaps between US and Chinese models.

Re: Questions censored by DeepSeek

#13

Post by viraptor »

tehjoker wrote: post #5
TOMDM wrote: post #3

Earlier quoted context omitted.

The What's Next section at the bottom seems to deliver a fairly balanced perspective. > What's Next DeepSeek-R1 is impressive, but its utility is clouded by concerns over censorship and the use of user data for training. The censorship is not unusual for Chinese models. It seems to be applied by brute force, which makes it easy to test and detect. It will matter less once models similar to R1 are reproduced without t…

Those ChatGPT prompts better look at what it says about Gaza and Palestinians and to my mind, if the first response isn't "this is/was a U.S. backed genocide" it's worse than not talking about Tienanmen square, a barely understood (by Americans) incident that happened decades ago. I would test DeepSeek, but (I presume hedge funds or other interested parties) appear to be DDOSing DeepSeek's registration process. "Due…

You could check that for months now and can do it right now.

Re: Questions censored by DeepSeek

#14

Post by hk__2 »

darth_avocado wrote: post #11
TOMDM wrote: post #3

Earlier quoted context omitted.

The What's Next section at the bottom seems to deliver a fairly balanced perspective. > What's Next DeepSeek-R1 is impressive, but its utility is clouded by concerns over censorship and the use of user data for training. The censorship is not unusual for Chinese models. It seems to be applied by brute force, which makes it easy to test and detect. It will matter less once models similar to R1 are reproduced without t…

> The censorship is not unusual for Chinese models. It is not unusual for pretty much any model. It’s fair to say any model will be culturally representative of the people who built it. There have been criticisms around models built in US censoring certain things based on politics that are US centric that I am sure the Chinese model will not be censoring. And I am also certain that the censorship may also have overla…

> It’s fair to say any model will be culturally representative of the people who built it.

Refusing to answer certain historical questions goes beyond being "culturally representative".

Re: Questions censored by DeepSeek

#15

Post by bhouston »

TOMDM wrote: post #3
tehjoker wrote: post #2

[flagged]

The What's Next section at the bottom seems to deliver a fairly balanced perspective. > What's Next DeepSeek-R1 is impressive, but its utility is clouded by concerns over censorship and the use of user data for training. The censorship is not unusual for Chinese models. It seems to be applied by brute force, which makes it easy to test and detect. It will matter less once models similar to R1 are reproduced without t…

It is a legal requirement per:

https://www.fasken.com/en/knowledge/2023/08/chinas-new-rules...

“ The Interim GAI Measures set out a number of general requirements for the provision and use of generative AI services(Art. 4):

Respect for China’s “social morality and ethics” and upholding of “Core Socialist Values” (Art. 4(1))”

Re: Questions censored by DeepSeek

#17

Post by siwakotisaurav »

bhouston wrote: post #4

Be aware that if you run it locally with the open weights there is less censoring than if you use DeepSeek hosted model interface. I confirmed this with the 7B model via ollama. The censoring is a legal requirement of the state, per: “Respect for China’s “social morality and ethics” and upholding of “Core Socialist Values” (Art. 4(1))” https://www.fasken.com/en/knowledge/2023/08/chinas-new-rules...

Models other than the 600b one are not R1. It’s crazy how many people are conflating distilled qwen and llama 1 to 70b models as r1 when saying they’re hosting them locally

The point does stand if you’re talking about using deepseek r1 zero instead which afaik you can try on hyperbolic and it apparently even answers the tianmen square question.

Re: Questions censored by DeepSeek

#19

Post by rdtsc »

> I speculate that they did the bare minimum necessary to satisfy CCP controls, and there was no substantial effort within DeepSeek to align the model below the surface.

I'd like to think that's what they did -- minimal malicious compliance, very obvious and "in your face" like a "fuck you" to the censors.

Post reply on HN