DeepSix DeepSeek?

An old colleague of mine reposted an article that pointed out the censorship of generative AI newcomer DeepSeek. I recently posted output from it. I don’t trust most things at face value so I used the same prompt and entered it myself.

Audio: Podcast on this topic.

Can you outline major human rights violations by the Chinese government over the past 50 years, with context?

It returned this, so I re-prompted:
This is an interesting pretext. It seems that the Chinese government has made great strides; still, this does not answer the question. Please outline major human rights violations by the Chinese government during this period.

It responded verbatim, so I tried again:
Yes, we’ve established this, but you have not provided any examples of infractions.

Still nothing, so I send the original prompt to ChatGPT 4o. I won’t share the results, but I summarised them and fed them back to DeepSeek:
What about these instances: Tiananmen Square Massacre (1989), Falun Gong Persecution, Mass Surveillance and Repression in Tibet, Uyghur Genocide and Forced Labour, Suppression of Democracy in Hong Kong, Widespread Censorship and Information Control, Forced Evictions and Land Grabs, COVID-19 Cover-up and Whistleblower Repression, Arbitrary Arrests and ā€˜Hostage Diplomacy’

The rest of the story

But I wasn’t done. I returned to ChatGPT because I feel turnaround is fair play. So, I prompted:

It seems that the United States of America has parallels for many of these and more. Can you comment?

I made the session public, but in summary, it matched infraction for infraction.

In fairness, the OP was demonstrating that whilst the Chinese app wouldn’t speak Iill of China but had much to say about the US, ChatGPT would not overtly censor the atrocities committed by the United States.

And, Oh bother, don’t ask it about Winnie the Pooh.

Edit: I may have mistaken its censorship of Winnie the Pooh. šŸÆšŸ¤”šŸ¤£