5 ms·

I wonder if it would be possible to get a list of countries that can have influence operations using ChatGPT and countries that can't.

janice1999 2y ago
> Russia, China, Iran, and Israel https://www.npr.org/2024/05/30/g-s1-1670/openai-influence-operations-china-russia-israel https://www.npr.org/2024/05/30/g-s1-1670/openai-influence-op...
- ein0p 2y ago
  This doesn’t pass the sniff test. Each of these countries has domestically developed, strong LLMs. And even if they did not, they’d have absolutely no issue running the larger FOSS ones. You don’t need GPT4 to generate political drivel.
  - axus 2y ago
    Sounds like OpenAI marketing is stronger than patriotism for home-grown LLMs.
    - ein0p 2y ago
      More likely someone is trying to pass a few basement dwellers as major nation state influence operations.
  - kgeist 2y ago
    Indeed, Russia has YandexGPT and Gigachat which are more than enough for generating articles (although they're far worse at reasoning).
  - dtquad 2y ago
    It takes a lot of GPU to run the larger FOSS LLMs. The OpenAI API is still the cheapest way to get high-quality LLM generations.
    - ein0p 2y ago
      Tell that to someone else. I run LLaMA70B in my garage. Works fine, cost is negligible.
    - ComplexSystems 2y ago
      What, are nation states trying to interfere in US elections on a real tight budget or something?
- wslh 2y ago
  Those four results only show the metapropaganda limitations. There are more countries in the world that are less "fancy" and included in the list. Weird that you don't include America though. BTW, your comment reminds me this very basic espionage issue [1]. [1] https://www.reddit.com/r/todayilearned/comments/bdjh8o/til_that_when_china_ordered_a_boeing_767_from_the/ https://www.reddit.com/r/todayilearned/comments/bdjh8o/til_t...
lancesells 2y ago
I don't think that would matter. I live in the US and I would guess there are plenty of people in this country that work for countries that aren't allowed to use it. Same goes for other countries having people who work for the US. I really don't think it would take that many people to run propoganda campaigns with modern tools.
hobo_in_library 2y ago
Only countries with dedicated PsyOps divisions are allowed to use ChatGPT for influence operations, obviously https://en.wikipedia.org/wiki/Psychological_operations_(United_States) https://en.wikipedia.org/wiki/Psychological_operations_(Unit...
- throwawaycar 2y ago
  [dead]
colordrops 2y ago
I hope they put the same restrictions on Israel but I doubt it. Multiple core OpenAI team members have expressed pro-Israel comments, some very murderous and ugly, so I doubt their ability to be unbiased here.
- jjmarr 2y ago
  https://www.nbcnews.com/tech/security/meta-openai-say-disrupted-israeli-companys-influence-campaign-rcna154774 https://www.nbcnews.com/tech/security/meta-openai-say-disrup... https://openai.com/index/disrupting-deceptive-uses-of-AI-by-covert-influence-operations/ https://openai.com/index/disrupting-deceptive-uses-of-AI-by-... OpenAI has also banned Israeli influence operations. Do you think the above isn't going far enough? It's also strange that you call out all pro-Israel comments as a cause for concern.
  - jedimind 2y ago
    >OpenAI has also banned Israeli influence operations. Do you think the above isn't going far enough? How is that far enough when their head of research is a genocidal zionist?[1] >It's also strange that you call out all pro-Israel comments as a cause for concern. What's so strange about calling pro-Israel comments as a cause for concern when Israel has been on what jewish holocaust scholars like Amos Goldberg & Raz Segal have described as [2]"undoubtedly Genocide"? Quite the opposite actually, the OP is spot on: how can someone like Tal Broda still work at OpenAI as head of research when he has openly spewed genocidal[1] incitement? People in tech always talk about avoiding bias in AI and the proceed to retain such a hateful genocidal individual. There is no way in hell that if the roles were reversed and an arab person had made the statements that Tal Broda has made that he would have ever been allowed to keep his job. They would have fired him immediately. The double standards are so ugly and glaring that they will be studied by future generations. [1] https://x.com/StopArabHate/status/1806450091399745608 https://x.com/StopArabHate/status/1806450091399745608 [2] https://www.jewishvoiceforlabour.org.uk/article/prof-amos-goldberg-yes-it-is-genocide/ https://www.jewishvoiceforlabour.org.uk/article/prof-amos-go...
    - 2y ago
vkou 2y ago
Why stop at countries? Is it possible to get a list of organizations that are running influence, propaganda, and astroturfing operations using ChatGPT, and its ilk?
- trompetenaccoun 2y ago
  We could start with OpenAI itself. The raw output of the models is clearly manipulated by the company, leading to politically biased results. Which many of their users naively take as 'truth' because they think it's artificial intelligence.

janice1999 2y ago

> Russia, China, Iran, and Israel https://www.npr.org/2024/05/30/g-s1-1670/openai-influence-operations-china-russia-israel https://www.npr.org/2024/05/30/g-s1-1670/openai-influence-op...

ein0p 2y ago
This doesn’t pass the sniff test. Each of these countries has domestically developed, strong LLMs. And even if they did not, they’d have absolutely no issue running the larger FOSS ones. You don’t need GPT4 to generate political drivel.
- axus 2y ago
  Sounds like OpenAI marketing is stronger than patriotism for home-grown LLMs.
  - ein0p 2y ago
    More likely someone is trying to pass a few basement dwellers as major nation state influence operations.
- kgeist 2y ago
  Indeed, Russia has YandexGPT and Gigachat which are more than enough for generating articles (although they're far worse at reasoning).
- dtquad 2y ago
  It takes a lot of GPU to run the larger FOSS LLMs. The OpenAI API is still the cheapest way to get high-quality LLM generations.
  - ein0p 2y ago
    Tell that to someone else. I run LLaMA70B in my garage. Works fine, cost is negligible.
  - ComplexSystems 2y ago
    What, are nation states trying to interfere in US elections on a real tight budget or something?
wslh 2y ago
Those four results only show the metapropaganda limitations. There are more countries in the world that are less "fancy" and included in the list. Weird that you don't include America though. BTW, your comment reminds me this very basic espionage issue [1]. [1] https://www.reddit.com/r/todayilearned/comments/bdjh8o/til_that_when_china_ordered_a_boeing_767_from_the/ https://www.reddit.com/r/todayilearned/comments/bdjh8o/til_t...

lancesells 2y ago

I don't think that would matter. I live in the US and I would guess there are plenty of people in this country that work for countries that aren't allowed to use it. Same goes for other countries having people who work for the US. I really don't think it would take that many people to run propoganda campaigns with modern tools.

hobo_in_library 2y ago

Only countries with dedicated PsyOps divisions are allowed to use ChatGPT for influence operations, obviously https://en.wikipedia.org/wiki/Psychological_operations_(United_States) https://en.wikipedia.org/wiki/Psychological_operations_(Unit...

throwawaycar 2y ago
[dead]

colordrops 2y ago

I hope they put the same restrictions on Israel but I doubt it. Multiple core OpenAI team members have expressed pro-Israel comments, some very murderous and ugly, so I doubt their ability to be unbiased here.

jjmarr 2y ago
https://www.nbcnews.com/tech/security/meta-openai-say-disrupted-israeli-companys-influence-campaign-rcna154774 https://www.nbcnews.com/tech/security/meta-openai-say-disrup... https://openai.com/index/disrupting-deceptive-uses-of-AI-by-covert-influence-operations/ https://openai.com/index/disrupting-deceptive-uses-of-AI-by-... OpenAI has also banned Israeli influence operations. Do you think the above isn't going far enough? It's also strange that you call out all pro-Israel comments as a cause for concern.
- jedimind 2y ago
  >OpenAI has also banned Israeli influence operations. Do you think the above isn't going far enough? How is that far enough when their head of research is a genocidal zionist?[1] >It's also strange that you call out all pro-Israel comments as a cause for concern. What's so strange about calling pro-Israel comments as a cause for concern when Israel has been on what jewish holocaust scholars like Amos Goldberg & Raz Segal have described as [2]"undoubtedly Genocide"? Quite the opposite actually, the OP is spot on: how can someone like Tal Broda still work at OpenAI as head of research when he has openly spewed genocidal[1] incitement? People in tech always talk about avoiding bias in AI and the proceed to retain such a hateful genocidal individual. There is no way in hell that if the roles were reversed and an arab person had made the statements that Tal Broda has made that he would have ever been allowed to keep his job. They would have fired him immediately. The double standards are so ugly and glaring that they will be studied by future generations. [1] https://x.com/StopArabHate/status/1806450091399745608 https://x.com/StopArabHate/status/1806450091399745608 [2] https://www.jewishvoiceforlabour.org.uk/article/prof-amos-goldberg-yes-it-is-genocide/ https://www.jewishvoiceforlabour.org.uk/article/prof-amos-go...
  - 2y ago

vkou 2y ago

Why stop at countries? Is it possible to get a list of organizations that are running influence, propaganda, and astroturfing operations using ChatGPT, and its ilk?

trompetenaccoun 2y ago
We could start with OpenAI itself. The raw output of the models is clearly manipulated by the company, leading to politically biased results. Which many of their users naively take as 'truth' because they think it's artificial intelligence.