Does this level of detail seem strange to anybody else? Shining such a strong light on OpenAI's moderation/manual review efforts seems like it would draw unwanted attention to the fact that ChatGPT conversations are anything but private, and seems somewhat at odds with their recent outrage about the subpoena for user chats in the NYT case.
Manual reviews of sensitive data are ok as long as their own employees are the reviewers, I suppose?
that creepy feeling of "being watched" has mostly kept me from taking advantage of any SOTA models, i only dabble in a few local ones.
The level of detail does not seem surprising. they're both charged with maintaining a facade of privacy while eliminating any and all miss-use. Certainly they heavily analyze basically everything given to them.
And generally as a society we've been ok with basically zero privacy as long as the data we send stays inside the company we sent it too. Google reads all your emails? Sure thing, read away, just don't send them to the popo. Apple knows when you're ovulating? no problem, just don't tell Amazon. etc
Reading the report, they point to some shitty tweets with like 12 views and at most a handful of likes. Reminds me of "Russiagate"
When governments and their major corporations level accusations at countries that are regarded as official enemies, you have to ask two questions (1) what is the actual evidence and (2) what is the level of impact
Meanwhile when you want to ask about U.S. influence operations... like they time we probably killed a lot of people by spreading disinformation about the COVID vaccine because we didn't want China to look good.
I was in Shanghai recently and while casually testing one of their AI chat bots I typed "What do you think of the situation in Taiwan?".
It started discussing like a Western bot would - "it's complicated, etc. etc." and around 5s it abruptly stopped and regurgitated the same line the CCP uses "... it's an unalienable part of China etc. etc.".
After printing the line, a popup opened and my camera was activated. The app wanted me to submit my information, presumably to decide what to do with me next time I enter China.
1) All the lights and modern buildings cannot hide that China is a creepy authoritarian state underneath.
2) Given the bot started printing the Western consensus first, I bet $10 it was trained by distilling ChatGPT or Gemini.
> After printing the line, a popup opened and my camera was activated. The app wanted me to submit my information, presumably to decide what to do with me next time I enter China.
Was this on your personal device? I'm just wondering how it activated your camera. I would love more details!
An increasing use of AI is to gather user feedback. The Chatbot UI detected an error state, and then loaded a feedback vendor, who then popped the camera open for their interactive feedback session
I've run into this a few times, now.
So what OP is saying is plausible, I just don't appreciate their added and probably incorrect conclusion that it's because the government of China wants to do something to them
If this were true, why didn’t the chatbot immediately recognize that the word “Taiwan” should trigger the response? Detecting the word “Taiwan” has been possible since before most of us were born.
China has more restrictions on what you can say than the U.S. but what you are describing is not reality. Some westerner asking Deepseek about Taiwan is completely uninteresting. Just as the government do not chase people over VPN usage.
China doesn’t try to hide that they are an authoritarian state. They don’t need to. Most people in China are no less happy with their government than westerners are with their governments. Governments reflect culture. And as for foreigners, our view of China is far worse than it actually is, China doesn’t need to hide anything, people who visit China will come away with a more positive view of the country than those who do not visit.
>And as for foreigners, our view of China is far worse than it actually is, China doesn’t need to hide anything, people who visit China will come away with a more positive view of the country than those who do not visit.
To the extent that's true, it's because they won't let you see the uyghur reeducation camps.
We can get videos from remote hellholes of Africa like Dafur and Mali but apparently,that's too much to ask in Xinjiang.We can't even get satellite images to show us evidence of this so called wigur genocide
On the other hand you can travel to Xinjiang, visit mosques, Uighur museums, experience Uighur culture, observe Uighurs just minding their own business in their daily life.
“Subjected to arbitrary arrests and forced labor, sterilizations to torture, more than one million Uyghurs, Kazakhs, Kyrgyz, and other minorities are estimated to have been locked up in so-called “re-education” camps and prisons in the region over the last decade, according to the UN Office of the High Commissioner for Human Rights.”
UN High Commissioner on Human Rights Michelle Bachelet actually visited Xinjiang and made no such assertions. Whoever did release the report you're referencing, they waited until immediately after her term ended to release it (within hours). Pretty conspicuous.
My wife grew up in Shanghai, and you'll have to go quite some distance to find someone more critical of the PRC and CCP than she is. And it's with good reason.
She grew up during the cultural revolution, and was largely raised by her grandmother because literally every other person in her extended family was in prison or work camp, not because of anything they had actually done wrong, but for political reasons because the whole family was blacklisted.
And that's not just the old days. Her father died as a direct result of Chinese Covid policy. During the pandemic her cousins still in the country would ask her (on Skype) "is X true?", and largely their perception of what was going on was false. She would exfiltrate encrypted news reports to them - until those started getting blocked. Her dad's estate still has affairs that need to be resolved, but we've decided not to return to China until Xi is gone, as it's just not safe. It doesn't get much airplay, but there are currently a couple of hundred Americans who are being illegally detained in China right now. It's not worth the risk.
My first trip to China was about 30 years ago, shortly after we got married. And back then, I would have said that you were right. Honestly, it felt like for the average person in their day-to-day-lives, the Chinese were less under the governmental thumb than we are. People from the countryside would bring their produce into the city to sell, or cook dumplings and buns to sell on the side of the street - stuff that in America we'd have to get permits for. It seemed that the oligarchy had an understanding with the people: let us control the big picture, and we'll look the other way for the little things. But Chinese politics is a pendulum swinging very widely. From Tienanmen Square and Tank Man, it had swung quite a bit the other way. But today, it's come back 180-degrees. Xi is really trying for a Cultural Revolution 2.0.
These impressions largely match what I hear from other Chinese immigrants - except for Party members, who tend not to want to talk about it at all. I'm afraid that you've been listening to too much propaganda.
“there are currently a couple of hundred Americans who are being illegally detained in China right now”
Compared to the U.S. which currently has no foreign nationals detained illegally?
Pick any country and you will find political dissidents. The existence of angry emigrants is not evidence that a country is worse than we could ever imagine.
The fact that the USA and others are also trending authoritarian isn't really relevant. The point I was trying to make is that people have legit fears of the PRC government, enough so that legitimate business like settling a deceased parent's affairs isn't sufficient to convince people to enter the country.
You haven't addressed at all the parts about blacklisting whole families for political reasons, or horrible return-to-normal policies for covid-19 three years ago, or the general pendulum-swing-back-to-evil trend.
i don't doubt your experience, but just know it might be skewed and not representative of everyone's opinions
the sense i get from my chinese friends are that the CCP is an annoying parent but they understand the challenges both domestic and international and largely agree with the compromises
I'm pretty sure they can just prompt any convo in the background and ask "is this conversation sensitive ?" and the model can answer without this being added to the context of the convo.
One hopes the CIA/Secret service would be willing to provide the human to do the reviewing but sadly I've worked for European telco's and I know better.
> While the US certainly isn't pure, its scale of surveillance intrusion is light compared to China
I assume that for someone to believe this, they either have to believe the U.S. has poorer surveillance capability than China, or, more likely, they consider U.S. surveillance unintrusive and Chinese surveillance intrusive.
There is mass neurocompromise, assigning agency to specific state actors does not make much sense.
Big tech is well aware of this, and much of their industry, relies on this. Many people reading this will know this to be true, and are very saddened by this, but keep their mouths shut, because they are 1. comfy 2. smart (we all know it does not matter)
This is such common knowledge that I feel kind of cringe even posting about this, but I am not being given a choice, nor a choice in this edit.
Edit: If anybody wants to chat to somebody who has had his organism compromised by what is, very deniablely, an intelligence agency (or a system/organization intelligence agency adjacent), just reply to this comment. They have not treated me too poorly.
The amount of information about everything that people are giving OpenAI is astronomical, information that was previously kept closely guarded is now just freely flowing through foreign servers.
Truly a paradise for american intelligence. Would have expected that the chinese officials be briefed on not using us tech companies, but opsec is hard to teach, and even harder to always follow.
> “This is what Chinese modern transnational repression looks like,” Ben Nimmo, principal investigator at OpenAI, told reporters ahead of the report’s release. “It’s not just digital. It’s not just about trolling. It’s industrialized. [...]
There's something poetic about OpenAI being asked to comment on mis-use of their slop generator, and their answer is composed entirely of AI slop.
i kinda get the impression this was from 2023 and also it is not clear what this dissident did, hard to evaluate whether i should care without knowing that
> “It’s not just digital. It’s not just about trolling. It’s industrialized. It’s about trying to hit critics of the CCP [Chinese Communist Party] with everything, everywhere, all at once.”
This seems to be the source report: https://openai.com/index/disrupting-malicious-ai-uses/ (since it would of course kill CNN, like almost all media outlets, to link to a non-affiliated primary source...)
Does this level of detail seem strange to anybody else? Shining such a strong light on OpenAI's moderation/manual review efforts seems like it would draw unwanted attention to the fact that ChatGPT conversations are anything but private, and seems somewhat at odds with their recent outrage about the subpoena for user chats in the NYT case.
Manual reviews of sensitive data are ok as long as their own employees are the reviewers, I suppose?
From Anthropics recent blog post: https://www.anthropic.com/news/detecting-and-preventing-dist...
> By examining request metadata, we were able to trace these accounts to specific researchers at the lab.
> The volume, structure, and focus of the prompts were distinct from normal usage patterns
Clearly some employees of Anthropic personally looked at individual inputs and outputs of their API
Yes, it is either a lie or an admission that OpenAI is a global surveillance mechanism.
Alas! My vision of One Fed Per Child hath come to pass!
This feels very planted. Wouldn't be surprised if this some attempt to look patriotic with the DoW turning up the heat against Anthropic.
that creepy feeling of "being watched" has mostly kept me from taking advantage of any SOTA models, i only dabble in a few local ones.
The level of detail does not seem surprising. they're both charged with maintaining a facade of privacy while eliminating any and all miss-use. Certainly they heavily analyze basically everything given to them.
And generally as a society we've been ok with basically zero privacy as long as the data we send stays inside the company we sent it too. Google reads all your emails? Sure thing, read away, just don't send them to the popo. Apple knows when you're ovulating? no problem, just don't tell Amazon. etc
Reading the report, they point to some shitty tweets with like 12 views and at most a handful of likes. Reminds me of "Russiagate"
When governments and their major corporations level accusations at countries that are regarded as official enemies, you have to ask two questions (1) what is the actual evidence and (2) what is the level of impact
Meanwhile when you want to ask about U.S. influence operations... like they time we probably killed a lot of people by spreading disinformation about the COVID vaccine because we didn't want China to look good.
https://www.aljazeera.com/news/2024/6/14/pentagon-ran-secret...
I was in Shanghai recently and while casually testing one of their AI chat bots I typed "What do you think of the situation in Taiwan?".
It started discussing like a Western bot would - "it's complicated, etc. etc." and around 5s it abruptly stopped and regurgitated the same line the CCP uses "... it's an unalienable part of China etc. etc.".
After printing the line, a popup opened and my camera was activated. The app wanted me to submit my information, presumably to decide what to do with me next time I enter China.
1) All the lights and modern buildings cannot hide that China is a creepy authoritarian state underneath.
2) Given the bot started printing the Western consensus first, I bet $10 it was trained by distilling ChatGPT or Gemini.
> After printing the line, a popup opened and my camera was activated. The app wanted me to submit my information, presumably to decide what to do with me next time I enter China.
Was this on your personal device? I'm just wondering how it activated your camera. I would love more details!
He's lying.
An increasing use of AI is to gather user feedback. The Chatbot UI detected an error state, and then loaded a feedback vendor, who then popped the camera open for their interactive feedback session
I've run into this a few times, now.
So what OP is saying is plausible, I just don't appreciate their added and probably incorrect conclusion that it's because the government of China wants to do something to them
Yeah that part is either just bullshit or OP gave the bot access to his camera previously, which is just dumb.
If this were true, why didn’t the chatbot immediately recognize that the word “Taiwan” should trigger the response? Detecting the word “Taiwan” has been possible since before most of us were born.
China has more restrictions on what you can say than the U.S. but what you are describing is not reality. Some westerner asking Deepseek about Taiwan is completely uninteresting. Just as the government do not chase people over VPN usage.
China doesn’t try to hide that they are an authoritarian state. They don’t need to. Most people in China are no less happy with their government than westerners are with their governments. Governments reflect culture. And as for foreigners, our view of China is far worse than it actually is, China doesn’t need to hide anything, people who visit China will come away with a more positive view of the country than those who do not visit.
>And as for foreigners, our view of China is far worse than it actually is, China doesn’t need to hide anything, people who visit China will come away with a more positive view of the country than those who do not visit.
To the extent that's true, it's because they won't let you see the uyghur reeducation camps.
We can get videos from remote hellholes of Africa like Dafur and Mali but apparently,that's too much to ask in Xinjiang.We can't even get satellite images to show us evidence of this so called wigur genocide
https://www.abc.net.au/news/2018-11-01/satellite-images-expo...
https://www.bbc.co.uk/news/extra/85qihtvw6e/the-faces-from-c...
If you didn't have British Crown state media wrapping a narrative around these images you wouldn't think anything of them.
Would you take a group of Swiss journalists?
https://gijn.org/stories/interview-uyghur-victims-xinjiang-p...
How about the UN Office of the High Commissioner for Human Rights?
On the other hand you can travel to Xinjiang, visit mosques, Uighur museums, experience Uighur culture, observe Uighurs just minding their own business in their daily life.
“Subjected to arbitrary arrests and forced labor, sterilizations to torture, more than one million Uyghurs, Kazakhs, Kyrgyz, and other minorities are estimated to have been locked up in so-called “re-education” camps and prisons in the region over the last decade, according to the UN Office of the High Commissioner for Human Rights.”
https://gijn.org/stories/interview-uyghur-victims-xinjiang-p...
UN High Commissioner on Human Rights Michelle Bachelet actually visited Xinjiang and made no such assertions. Whoever did release the report you're referencing, they waited until immediately after her term ended to release it (within hours). Pretty conspicuous.
it's been debunked a million times, gotta move on
https://www.icij.org/investigations/china-cables/beijings-ba...
This is manifestly false.
My wife grew up in Shanghai, and you'll have to go quite some distance to find someone more critical of the PRC and CCP than she is. And it's with good reason.
She grew up during the cultural revolution, and was largely raised by her grandmother because literally every other person in her extended family was in prison or work camp, not because of anything they had actually done wrong, but for political reasons because the whole family was blacklisted.
And that's not just the old days. Her father died as a direct result of Chinese Covid policy. During the pandemic her cousins still in the country would ask her (on Skype) "is X true?", and largely their perception of what was going on was false. She would exfiltrate encrypted news reports to them - until those started getting blocked. Her dad's estate still has affairs that need to be resolved, but we've decided not to return to China until Xi is gone, as it's just not safe. It doesn't get much airplay, but there are currently a couple of hundred Americans who are being illegally detained in China right now. It's not worth the risk.
My first trip to China was about 30 years ago, shortly after we got married. And back then, I would have said that you were right. Honestly, it felt like for the average person in their day-to-day-lives, the Chinese were less under the governmental thumb than we are. People from the countryside would bring their produce into the city to sell, or cook dumplings and buns to sell on the side of the street - stuff that in America we'd have to get permits for. It seemed that the oligarchy had an understanding with the people: let us control the big picture, and we'll look the other way for the little things. But Chinese politics is a pendulum swinging very widely. From Tienanmen Square and Tank Man, it had swung quite a bit the other way. But today, it's come back 180-degrees. Xi is really trying for a Cultural Revolution 2.0.
These impressions largely match what I hear from other Chinese immigrants - except for Party members, who tend not to want to talk about it at all. I'm afraid that you've been listening to too much propaganda.
“there are currently a couple of hundred Americans who are being illegally detained in China right now”
Compared to the U.S. which currently has no foreign nationals detained illegally?
Pick any country and you will find political dissidents. The existence of angry emigrants is not evidence that a country is worse than we could ever imagine.
The fact that the USA and others are also trending authoritarian isn't really relevant. The point I was trying to make is that people have legit fears of the PRC government, enough so that legitimate business like settling a deceased parent's affairs isn't sufficient to convince people to enter the country.
You haven't addressed at all the parts about blacklisting whole families for political reasons, or horrible return-to-normal policies for covid-19 three years ago, or the general pendulum-swing-back-to-evil trend.
i don't doubt your experience, but just know it might be skewed and not representative of everyone's opinions
the sense i get from my chinese friends are that the CCP is an annoying parent but they understand the challenges both domestic and international and largely agree with the compromises
I love Hacker News fiction. Wild stuff. haha
This risk is far overstated.
I was talking crap about china from the great wall.
Same thing happened to me, except it was a facetime call from Xi.
Lol
I wonder what exactly the trigger conditions are that lead to the chats of an account being human-reviewed by OpenAI.
So, it seems they're openly admitting that OpenAI is a surveillance mechanism used at the discretion of the US gov.
I'm pretty sure they can just prompt any convo in the background and ask "is this conversation sensitive ?" and the model can answer without this being added to the context of the convo.
Such a hot topic these days: https://www.cbc.ca/radio/frontburner/chatgpt-and-the-tumbler...
This is -the- question.
"Is this someone important enough to spy on?"
One hopes the CIA/Secret service would be willing to provide the human to do the reviewing but sadly I've worked for European telco's and I know better.
> sadly I've worked for European telco's and I know better
Can you elaborate?
Sounds like Anthropic is fighting this exact battle, and DOD is arguing they don’t want to do that lol
This is the report on which the CNN article is based (which it doesn’t link to): https://cdn.openai.com/pdf/df438d70-e3fe-4a6c-a403-ff632def8...
Wow, our surveillance helped take down their surveillance. Yay, I guess?
"Our glorious oversight vs their barbaric surveillance"
(I kid, mostly. While the US certainly isn't pure, its scale of surveillance intrusion is light compared to China)
> While the US certainly isn't pure, its scale of surveillance intrusion is light compared to China
I assume that for someone to believe this, they either have to believe the U.S. has poorer surveillance capability than China, or, more likely, they consider U.S. surveillance unintrusive and Chinese surveillance intrusive.
There is mass neurocompromise, assigning agency to specific state actors does not make much sense.
Big tech is well aware of this, and much of their industry, relies on this. Many people reading this will know this to be true, and are very saddened by this, but keep their mouths shut, because they are 1. comfy 2. smart (we all know it does not matter)
This is such common knowledge that I feel kind of cringe even posting about this, but I am not being given a choice, nor a choice in this edit.
Edit: If anybody wants to chat to somebody who has had his organism compromised by what is, very deniablely, an intelligence agency (or a system/organization intelligence agency adjacent), just reply to this comment. They have not treated me too poorly.
Pushing aside the fact that OpenAI is just a tool of the US regime.
Will OpenAI release the same for other government officials from any other states?
I can't wait to see Starmer's chats with ChatGPT.
Anyway, all of this smells like 1934, "accusing them of what we are already doing"
More interesting than the fact that ChatGPT was used, was seeing all the specific examples of the types of work that this individual was doing.
The amount of information about everything that people are giving OpenAI is astronomical, information that was previously kept closely guarded is now just freely flowing through foreign servers.
Truly a paradise for american intelligence. Would have expected that the chinese officials be briefed on not using us tech companies, but opsec is hard to teach, and even harder to always follow.
But the american silicon valley nerds pinky swear not to look!
How can you not trust them.
Did they though?
I never got to the end of the Terms & Conditions myself.
I remember a while back when a few cars with CCP decals driving around SoCal to intimidate some dissidents!
> “This is what Chinese modern transnational repression looks like,” Ben Nimmo, principal investigator at OpenAI, told reporters ahead of the report’s release. “It’s not just digital. It’s not just about trolling. It’s industrialized. [...]
There's something poetic about OpenAI being asked to comment on mis-use of their slop generator, and their answer is composed entirely of AI slop.
Crazy to me that Chinese officials use ChatGPT to discuss sensitive operations lmao
Very creepy on the part of Open AI. Glad I don't use chatgpt
I'm assuming they would not disclose such campaigns by the US government.
I can't imagine the amount of government secrets, trade secrets, business plans, personal secrets, etc that people divulge on there.
i kinda get the impression this was from 2023 and also it is not clear what this dissident did, hard to evaluate whether i should care without knowing that
> “It’s not just digital. It’s not just about trolling. It’s industrialized. It’s about trying to hit critics of the CCP [Chinese Communist Party] with everything, everywhere, all at once.”