🚩 Report: Legal issue(s)
I am an AI developer with personal experience dealing with data privacy and collection/storage laws. This is not a legally collected or published dataset. The data is public but only can be collected with user consent or ability to opt out. There is no GDPR or CCPA opt out ability, data removal ability, and there is zero anonymization to redact personal information.
Publically available does not mean legal to copy and post in any format for any reason. YouTube videos are public, but copyrighted and and privately owned. Street signs are publically available, but you can't saw one off and take it home as decoration. There are legitimate laws being broken here, and you would be wise to at least attempt to comply with them to avoid dataset removal or even litigation.
This kind of behavior is overall harmful to the ML field, since by your bluesky feed its clear you are intentionally trying to troll or even harass the very same people who you would rely on to use your services in the future, and it does the entire industry a disservice to behave in this manner. People like you who are clearly trying to stir people up with complete disregard for consent or even your own public image honestly make me ashamed to work in this field. I hope you learn some common sense and stop embarrassing yourself, your company, and the industry.
https://commission.europa.eu/law/law-topic/data-protection/reform/what-personal-data_en
It depends if the data collected is anonymized or not. I am not qualified to tell if the PID can be trace back to an identifiable person. If it's not, it's conform to GDPR, if it is, the dataset owner just needs to drop/remove the user column.
https://commission.europa.eu/law/law-topic/data-protection/reform/what-personal-data_en
It depends if the data collected is anonymized or not. I am not qualified to tell if the PID can be trace back to an identifiable person. If it's not, it's conform to GDPR, if it is, the dataset owner just needs to drop/remove the user column.
the dataset contains links to user handles which is clearly not anonymized, since they link directly to individual users. Even if it didn't, there's still a risk for CCPA/GDPR violations because the content of the text can contain personal information that was not consented to be shared on this website. It was consented to be shared on AtProto websites, which have actual GDPR/CCPA processes, unlike random AI developers who have a nightlife of acting like an internet gremlin.
https://commission.europa.eu/law/law-topic/data-protection/reform/what-personal-data_en
It depends if the data collected is anonymized or not. I am not qualified to tell if the PID can be trace back to an identifiable person. If it's not, it's conform to GDPR, if it is, the dataset owner just needs to drop/remove the user column.the dataset contains links to user handles which is clearly not anonymized, since they link directly to individual users. Even if it didn't, there's still a risk for CCPA/GDPR violations because the content of the text can contain personal information that was not consented to be shared on this website. It was consented to be shared on AtProto websites, which have actual GDPR/CCPA processes, unlike random AI developers who have a nightlife of acting like an internet gremlin.
Good point, thanks for the precision on the actual text content.
https://commission.europa.eu/law/law-topic/data-protection/reform/what-personal-data_en
It depends if the data collected is anonymized or not. I am not qualified to tell if the PID can be trace back to an identifiable person. If it's not, it's conform to GDPR, if it is, the dataset owner just needs to drop/remove the user column.the dataset contains links to user handles which is clearly not anonymized, since they link directly to individual users. Even if it didn't, there's still a risk for CCPA/GDPR violations because the content of the text can contain personal information that was not consented to be shared on this website. It was consented to be shared on AtProto websites, which have actual GDPR/CCPA processes, unlike random AI developers who have a nightlife of acting like an internet gremlin.
Good point, thanks for the precision on the actual text content.
Companies that deal with EU citizens can't even keep spreadsheets of data from their services on their work computer with personal information because it could become a GDPR violation, and no one even sees those. Publishing this stuff online is a massive violation in many ways. These are serious laws with very serious ramifications for violations.
I am an AI developer with personal experience dealing with data privacy and collection/storage laws. This is not a legally collected or published dataset. The data is public but only can be collected with user consent or ability to opt out. There is no GDPR or CCPA opt out ability, data removal ability, and there is zero anonymization to redact personal information.
Publically available does not mean legal to copy and post in any format for any reason. YouTube videos are public, but copyrighted and and privately owned. Street signs are publically available, but you can't saw one off and take it home as decoration. There are legitimate laws being broken here, and you would be wise to at least attempt to comply with them to avoid dataset removal or even litigation.
This kind of behavior is overall harmful to the ML field, since by your bluesky feed its clear you are intentionally trying to troll or even harass the very same people who you would rely on to use your services in the future, and it does the entire industry a disservice to behave in this manner. People like you who are clearly trying to stir people up with complete disregard for consent or even your own public image honestly make me ashamed to work in this field. I hope you learn some common sense and stop embarrassing yourself, your company, and the industry.
Legal where?
I am an AI developer with personal experience dealing with data privacy and collection/storage laws. This is not a legally collected or published dataset. The data is public but only can be collected with user consent or ability to opt out. There is no GDPR or CCPA opt out ability, data removal ability, and there is zero anonymization to redact personal information.
Publically available does not mean legal to copy and post in any format for any reason. YouTube videos are public, but copyrighted and and privately owned. Street signs are publically available, but you can't saw one off and take it home as decoration. There are legitimate laws being broken here, and you would be wise to at least attempt to comply with them to avoid dataset removal or even litigation.
This kind of behavior is overall harmful to the ML field, since by your bluesky feed its clear you are intentionally trying to troll or even harass the very same people who you would rely on to use your services in the future, and it does the entire industry a disservice to behave in this manner. People like you who are clearly trying to stir people up with complete disregard for consent or even your own public image honestly make me ashamed to work in this field. I hope you learn some common sense and stop embarrassing yourself, your company, and the industry.
Legal where?
It does not matter where you choose to commit data privacy law violations from, you are still liable for them if the people whose data you illegally use live in california or the EU. Crimes and civil liability are decided based on where the victim resides, not the offender. Unless you live in a country that will protect you like Russia, good luck.
You seem to have a lot of time on your hands for an "AI developer" (or whatever buzzword this thing is). I'm quite sure most of the pretrain models you're using are trained on copyrighted data anyway. Must feel good to be the internet white knight.
You seem to have a lot of time on your hands for an "AI developer" (or whatever buzzword this thing is). I'm quite sure most of the pretrain models you're using are trained on copyrighted data anyway. Must feel good to be the internet white knight.
In case you didn't notice, this week is Thanksgiving so I do indeed have more time to call out blatant law violations and self-sabatoging trolls who are destroying the reputation of AI. I'm sorry you're jealous I actually have a career in the field, but people who are serious about AI care about things like PR and reputation, which the children who have overtaken this forum certainly don't.
And nah, I don't mess around with LLMs lol.
It seems like your surface knowledge is limited to LLMs, while almost all transformer models are trained on various degrees of copyrighted data, at least you learned something now. AI is the new crypto, everyone and their dog likes to brag about it and I obviously don't care nor asked. Making a drama here won't earn you any goodboy points. Have a nice day
It seems like your surface knowledge is limited to LLMs, while almost all transformer models are trained on various degrees of copyrighted data, at least you learned something now. AI is the new crypto, everyone and their dog likes to brag about it and I obviously don't care nor asked. Making a drama here won't earn you any goodboy points. Have a nice day
it seems like you still have no idea what you're talking about. The issue here isn't even copyright, it's lack of consent for personal data published on the internet, without any opt out or deletion mechanisms. It doesn't matter what the data is intended to be used for, the publishing of the data is the issue. And no, models trained on copyrighted data that aren't generative in nature have no copyright issues since they can't replicate the data. Only generative models can, which is why I mentioned them. You aren't half as smart as you think.
And no, I didn't start working on AI yesterday. I worked on ML for many years before the recent hype brought in bad and self destructive unethical actors like you and OP who are ruining the field and poisoning the well.
You all are so butthurt it's hilarious. Don't post things publicly if you don't want them saved. Also, people are in the US. One can simply geoblock this dataset from the EU.
You all are so butthurt it's hilarious. Don't post things publicly if you don't want them saved. Also, people are in the US. One can simply geoblock this dataset from the EU.
Your personal opinion is not relevant to what’s legal or not.
couldn't someone report this repository as a GDPR violation?
couldn't someone report this repository as a GDPR violation?
Lol, GDPR violation. Just geoblock EU and that issue is simply gone.
there's nothing illegal with scraping data when the person scraper has never agreed to any TOS, just like a human can view it, a scraper can view it, and if you don't want a scraper to view it, then you why are you ok with any human viewing it? what's public is public
Street signs are publically available, but you can't saw one off and take it home as decoration.
not the same, duplicating the data does not DESTROY the original data. If I had a magic spell that could duplicate a traffic light and take it home, then you definitely cannot tell me to not duplicate the traffic light and take it home.
there's nothing illegal with scraping data when the person scraper has never agreed to any TOS, just like a human can view it, a scraper can view it, and if you don't want a scraper to view it, then you why are you ok with any human viewing it? what's public is public
Scraping, hosting and trafficking personal data without explicit consent is in fact illegal in many countries. It's illegal in the EU (where HF is legally established), and it applies to data of EU citizens regardless of where the data is stored.
"You can read on for more detail. Please keep in mind that because the AT Protocol is decentralized, Developer Applications are operated by third parties (“Developers”) who Bluesky doesn’t control and who may have their own privacy practices. Developer Applications may have their own Privacy Notices, which we encourage you to read."
@GuillaumeBHF
"You can read on for more detail. Please keep in mind that because the AT Protocol is decentralized, Developer Applications are operated by third parties (“Developers”) who Bluesky doesn’t control and who may have their own privacy practices. Developer Applications may have their own Privacy Notices, which we encourage you to read."
@GuillaumeBHF
My comment above still applies. The law applies regardless.
"You can read on for more detail. Please keep in mind that because the AT Protocol is decentralized, Developer Applications are operated by third parties (“Developers”) who Bluesky doesn’t control and who may have their own privacy practices. Developer Applications may have their own Privacy Notices, which we encourage you to read."
@GuillaumeBHFMy comment above still applies. The law applies regardless.
okay well than its on you to file/contact the eu and have them take action. you know how many data privacy violations go on daily? change only happens with action
okay well than its on you to file/contact the eu and have them take action.
Legally, the first step is to ask whoever is publishing the data to remove them, which is what people have been doing here.
Then if nothing happens, ask the hosting provider: people have been emailing HF and trying to get in touch with them by various means.
In most cases either of those cooperate and the problem is solved.
If not then the national Data Protection Authority gets a signalment, investigates, and issues fines & order content removal. HF is a multi-billion dollar company promoting responsible open source research and has an important role in handling this better, hence our attempts to get them to do something first. Also their code of conduct covers this type of issues. But you're right about the next steps if nothing happens.
you know how many data privacy violations go on daily? change only happens with action
That's exactly why these laws have been enacted.
TBH my original complaint here wasn't intended to be a public debate, but this is the only space to flag issues with the repo and inform the owner AFAIK.
But then I was surprised that so many people here were replying with completely wrong assumptions about the legal and ethical aspects of handling personal data, which is why I am taking the time to reply.
Open source is great, I am all for responsible AI. But let's not reproduce in the OS AI community the predatory practices some big closed-sourced companies use. We can do better.
Street signs are publically available, but you can't saw one off and take it home as decoration.
not the same, duplicating the data does not DESTROY the original data. If I had a magic spell that could duplicate a traffic light and take it home, then you definitely cannot tell me to not duplicate the traffic light and take it home.
I like how you ignored the other analogy (stealing youtube videos) because it was inconvenient to your argument. The point is being public does not mean its your property to do anything with. Data is considered property, even though it can be copied, despite what AI bros think. That's the entire reason we have copyright and data privacy laws to begin with.
okay well than its on you to file/contact the eu and have them take action.
Legally, the first step is to ask whoever is publishing the data to remove them, which is what people have been doing here.
Then if nothing happens, ask the hosting provider: people have been emailing HF and trying to get in touch with them by various means.
In most cases either of those cooperate and the problem is solved.If not then the national Data Protection Authority gets a signalment, investigates, and issues fines & order content removal. HF is a multi-billion dollar company promoting responsible open source research and has an important role in handling this better, hence our attempts to get them to do something first. Also their code of conduct covers this type of issues. But you're right about the next steps if nothing happens.
you know how many data privacy violations go on daily? change only happens with action
That's exactly why these laws have been enacted.
TBH my original complaint here wasn't intended to be a public debate, but this is the only space to flag issues with the repo and inform the owner AFAIK.
But then I was surprised that so many people here were replying with completely wrong assumptions about the legal and ethical aspects of handling personal data, which is why I am taking the time to reply.Open source is great, I am all for responsible AI. But let's not reproduce in the OS AI community the predatory practices some big closed-sourced companies use. We can do better.
Yup, a lot of people are filing complaints with CA and EU due to the stated intention to continue violating privacy laws. I hope alpin and pygmalion enjoy the lawyers fees!
Street signs are publically available, but you can't saw one off and take it home as decoration.
not the same, duplicating the data does not DESTROY the original data. If I had a magic spell that could duplicate a traffic light and take it home, then you definitely cannot tell me to not duplicate the traffic light and take it home.
I like how you ignored the other analogy (stealing youtube videos) because it was inconvenient to your argument. The point is being public does not mean its your property to do anything with. Data is considered property, even though it can be copied, despite what AI bros think. That's the entire reason we have copyright and data privacy laws to begin with.
Street sign is obvs gov property. but by using blue sky your are giving permission for the data to be used like this, its free. Im not sure how EU data laws are but if bluesky doesnt have any clause for user privacy than I dont think a case can really be made. Since you read and agreeded to Blueskys TOS
Street signs are publically available, but you can't saw one off and take it home as decoration.
not the same, duplicating the data does not DESTROY the original data. If I had a magic spell that could duplicate a traffic light and take it home, then you definitely cannot tell me to not duplicate the traffic light and take it home.
I like how you ignored the other analogy (stealing youtube videos) because it was inconvenient to your argument. The point is being public does not mean its your property to do anything with. Data is considered property, even though it can be copied, despite what AI bros think. That's the entire reason we have copyright and data privacy laws to begin with.
Street sign is obvs gov property. but by using blue sky your are giving permission for the data to be used like this, its free. Im not sure how EU data laws are but if bluesky doesnt have any clause for user privacy than I dont think a case can really be made. Since you read and agreeded to Blueskys TOS
According to section 2C of Bluesky's ToS, we are only giving permission for our posts to be distributed either on Bluesky's website or via the AT Protocol. HuggingFace is neither Bluesky nor distributing these datasets via the AT Protocol so that does not protect them here.
Bad news bears. The AT Protocol is exactly what distributed these posts. Keep grasping at straws though. Last I checked, TOS also weren't laws. If there is a problem regarding the terms, that's between bluesky and whatever account was using the site. IIRC, the most they can do is ban you.
Street signs are publically available, but you can't saw one off and take it home as decoration.
not the same, duplicating the data does not DESTROY the original data. If I had a magic spell that could duplicate a traffic light and take it home, then you definitely cannot tell me to not duplicate the traffic light and take it home.
I like how you ignored the other analogy (stealing youtube videos) because it was inconvenient to your argument. The point is being public does not mean its your property to do anything with. Data is considered property, even though it can be copied, despite what AI bros think. That's the entire reason we have copyright and data privacy laws to begin with.
Street sign is obvs gov property. but by using blue sky your are giving permission for the data to be used like this, its free. Im not sure how EU data laws are but if bluesky doesnt have any clause for user privacy than I dont think a case can really be made. Since you read and agreeded to Blueskys TOS
According to section 2C of Bluesky's ToS, we are only giving permission for our posts to be distributed either on Bluesky's website or via the AT Protocol. HuggingFace is neither Bluesky nor distributing these datasets via the AT Protocol so that does not protect them here.
The owner of this dataset used AT to collect this data, they did not scrape bluesky. Is there something im missing or are you farting out your rear a bit?
Street signs are publically available, but you can't saw one off and take it home as decoration.
not the same, duplicating the data does not DESTROY the original data. If I had a magic spell that could duplicate a traffic light and take it home, then you definitely cannot tell me to not duplicate the traffic light and take it home.
I like how you ignored the other analogy (stealing youtube videos) because it was inconvenient to your argument. The point is being public does not mean its your property to do anything with. Data is considered property, even though it can be copied, despite what AI bros think. That's the entire reason we have copyright and data privacy laws to begin with.
Street sign is obvs gov property. but by using blue sky your are giving permission for the data to be used like this, its free. Im not sure how EU data laws are but if bluesky doesnt have any clause for user privacy than I dont think a case can really be made. Since you read and agreeded to Blueskys TOS
According to section 2C of Bluesky's ToS, we are only giving permission for our posts to be distributed either on Bluesky's website or via the AT Protocol. HuggingFace is neither Bluesky nor distributing these datasets via the AT Protocol so that does not protect them here.
The owner of this dataset used AT to collect this data, they did not scrape bluesky. Is there something im missing or are you farting out your rear a bit?
From this dataset's information card:
This dataset contains 2 million public posts collected from Bluesky Social's firehose API
Street signs are publically available, but you can't saw one off and take it home as decoration.
not the same, duplicating the data does not DESTROY the original data. If I had a magic spell that could duplicate a traffic light and take it home, then you definitely cannot tell me to not duplicate the traffic light and take it home.
I like how you ignored the other analogy (stealing youtube videos) because it was inconvenient to your argument. The point is being public does not mean its your property to do anything with. Data is considered property, even though it can be copied, despite what AI bros think. That's the entire reason we have copyright and data privacy laws to begin with.
Street sign is obvs gov property. but by using blue sky your are giving permission for the data to be used like this, its free. Im not sure how EU data laws are but if bluesky doesnt have any clause for user privacy than I dont think a case can really be made. Since you read and agreeded to Blueskys TOS
According to section 2C of Bluesky's ToS, we are only giving permission for our posts to be distributed either on Bluesky's website or via the AT Protocol. HuggingFace is neither Bluesky nor distributing these datasets via the AT Protocol so that does not protect them here.
The owner of this dataset used AT to collect this data, they did not scrape bluesky. Is there something im missing or are you farting out your rear a bit?
From this dataset's information card:
This dataset contains 2 million public posts collected from Bluesky Social's firehose API
Im coming to your house btw
According to section 2C of Bluesky's ToS, we are only giving permission for our posts to be distributed either on Bluesky's website or via the AT Protocol.
Actually, it is more conservative than that. Users are ONLY giving permissions to Bluesky, PBC via that ToS. This ToS does not say anything about other entities using the AT Protocol. Let's look at 2C.
Bluesky’s Use of User Content: User Content is only used only in connection with: (a) providing Bluesky Social and Content, including by sharing User Content throughout Bluesky Social and the AT Protocol as well as developing and improving our current and future offerings; and (b) promoting and marketing Bluesky and Bluesky Social.
This item is for "Bluesky's Use". The subject is "Bluesky". And ToS also say:
When we say “Bluesky,” “we,” “us,” and “our” in these Terms, we mean Bluesky, PBC.
So, users did NOT give permission to other third parties via the AT Protocol by accepting that ToS.
Let's talk about the AT Protocol. It is a protocol that allows users to access data publically. However, it does not mean that anybody can use the data in whatever way they want to use.
- For example, there are many publically accessible websites via HTTP. However, many of these websites have copyright marks or Creative Contents licenses. Anyone who uses the contents are expected to follow such licenses. And, a work can still be protected by copyright even if it doesn't have a visible copyright mark.
- For another example, let's think about software licenses. GNU license prevents a private company can copy the code and use it for profit, even though anybody can publically access the code from GitHub.
So, "publically accessible" does not necessarily mean that users gave permission to use their contents to anybody.
Getting consent is important for AI developers too to protect themselves - theoretically, if a model is trained from dataset that has a legal issue, all the derived data, including the model itself, should be deleted. Tracking data dependencies and retraining models are expensive and annoying. It's better to start from clean and safe dataset.
It seems like most AI people in HF already agree Bluesky data is not that useful anyway. This is just becoming trolling, making normal Bluesky users angry and concerned, as well as jeopardizing HF - I already see many privacy concerned people reporting this incident regarding GDPR violation, and HF might need to take actions (or get fined).
Let's talk about the AT Protocol. It is a protocol that allows users to access data publically. However, it does not mean that anybody can use the data in whatever way they want to use.
- For example, there are many publically accessible websites via HTTP. However, many of these websites have copyright marks or Creative Contents licenses. Anyone who uses the contents are expected to follow such licenses. And, a work can still be protected by copyright even if it doesn't have a visible copyright mark.
- For another example, let's think about software licenses. GNU license prevents a private company can copy the code and use it for profit, even though anybody can publically access the code from GitHub.
So, "publically accessible" does not necessarily mean that users gave permission to use their contents to anybody.
Getting consent is important for AI developers too to protect themselves - theoretically, if a model is trained from dataset that has a legal issue, all the derived data, including the model itself, should be deleted. Tracking data dependencies and retraining models are expensive and annoying. It's better to start from clean and safe dataset.
It seems like most AI people in HF already agree Bluesky data is not that useful anyway. This is just becoming trolling, making normal Bluesky users angry and concerned, as well as jeopardizing HF - I already see many privacy concerned people reporting this incident regarding GDPR violation, and HF might need to take actions (or get fined).
you give consent by using the platform. Fight for better digital rights.
According to section 2C of Bluesky's ToS, we are only giving permission for our posts to be distributed either on Bluesky's website or via the AT Protocol.
Actually, it is more conservative than that. Users are ONLY giving permissions to Bluesky, PBC via that ToS. This ToS does not say anything about other entities using the AT Protocol. Let's look at 2C.
Bluesky’s Use of User Content: User Content is only used only in connection with: (a) providing Bluesky Social and Content, including by sharing User Content throughout Bluesky Social and the AT Protocol as well as developing and improving our current and future offerings; and (b) promoting and marketing Bluesky and Bluesky Social.
This item is for "Bluesky's Use". The subject is "Bluesky". And ToS also say:
When we say “Bluesky,” “we,” “us,” and “our” in these Terms, we mean Bluesky, PBC.
So, users did NOT give permission to other third parties via the AT Protocol by accepting that ToS.
okay so ignoring 2B ?
@nostopasking Yes, as you pointed out, it is user's responsibility. Bluesky, PBC will not and cannot sue HF+alpindale to protect users. I am just saying that the users did not give permission to HF+alpindale for usage by that ToS.
What users can do? At least I know several people already reported to the DPA for GDPR violation. Let the whatever authorities decide if this is okay or not, if HF+alpindale do want to keep doing this, ignoring the user's requests.
But do not use ToS to justify what alpindale does here. It is just irrelevant. (Same for AT protocol and Firehose.)
you give consent by using the platform. Fight for better digital rights.
Cute. Where in the ToS says that?
It is even cuter (actually appalling) that you think keeping the scraped data here is "fighting for better digital rights".
you give consent by using the platform. Fight for better digital rights.
Cute. Where in the ToS says that?
It is even cuter (actually appalling) that you think keeping the scraped data here is "fighting for better digital rights".
ToS is posted above, also its not scraped as you yourself said. And I said "Fight for better digital rights", maybe it should be on companies to have clearer warnings/disclaimers about their products like FDA requires for food. Seems that these days people in general arent as concerned as they should be with how their data is used.
"you give consent by using the platform"
Have you graduated Grade 12? the ToS is pretty clear. But lmk if you need me to reword anything or dont understand the content.
"Bluesky's Use". "When we say “Bluesky,” “we,” “us,” and “our” in these Terms, we mean Bluesky, PBC."
Not HF.
@nostopasking Hey, if you want to believe what you want to believe, do it. I do not care any more. I am sure other people in this thread would feel the same.
why do people even bother defending things like this dataset? it's a clear violation of privacy laws and morals. how would you react if something like twitter took everyone's tweets and compiled a massive dataset with attachments, names, dates, replies, likes, retweets, etc. and released it to the public?
ignoring the legal side for a moment, if anyone edited or deleted something after the dataset was collected, then it would remain as it was before in the dataset. what if someone posted a private link, image, or other personal info on accident, the dataset caught that, no measures were put in place to censor or remove that data, and twitter refused to remove the data after the fact?
that's exactly what's happening here in this dataset.
why do people even bother defending things like this dataset? it's a clear violation of privacy laws and morals. how would you react if something like twitter took everyone's tweets and compiled a massive dataset with attachments, names, dates, replies, likes, retweets, etc. and released it to the public?
Because all the people who are defending this have absolutely zero understanding of how the law works or even how to properly read a legal document, as is proven by posts above.
why do people even bother defending things like this dataset? it's a clear violation of privacy laws and morals. how would you react if something like twitter took everyone's tweets and compiled a massive dataset with attachments, names, dates, replies, likes, retweets, etc. and released it to the public?
ignoring the legal side for a moment, if anyone edited or deleted something after the dataset was collected, then it would remain as it was before in the dataset. what if someone posted a private link, image, or other personal info on accident, the dataset caught that, no measures were put in place to censor or remove that data, and twitter refused to remove the data after the fact?
that's exactly what's happening here in this dataset.
They do already it just not public access you can pay for it though. If something was deleated after the fact than I should have been responsible enough to not post whatever it was, although deleting is subjective and most sites will still store your userdata (whatever was 'deleted').
why do people even bother defending things like this dataset? it's a clear violation of privacy laws and morals. how would you react if something like twitter took everyone's tweets and compiled a massive dataset with attachments, names, dates, replies, likes, retweets, etc. and released it to the public?
Because all the people who are defending this have absolutely zero understanding of how the law works or even how to properly read a legal document, as is proven by posts above.
Hey, id like to know where my misunderstanding is. I read most ToS documents before using platforms so I thought I had a pretty good grasp. Maybe im really off the mark so it would be great if youd lmk!
They do already it just not public access you can pay for it though. If something was deleated after the fact than I should have been responsible enough to not post whatever it was, although deleting is subjective and most sites will still store your userdata (whatever was 'deleted').
It might be legal in the US, it's illegal in Europe in most cases, and users have a legal right to request (and obtain) their data to be removed from such datasets. Please check the GDPR laws. People here are simply demanding that these laws be respected, and you come tell them to swallow it and move on. Don't be that person.
Regardless of current laws, and how/where they apply, Open Source AI can and should encourage fairer practices than large closed-sourced tech companies.
Please read the HF ToS, Code of Conduct and Privacy Policy, you keep missing the mark on what's acceptable.
why do people even bother defending things like this dataset? it's a clear violation of privacy laws and morals. how would you react if something like twitter took everyone's tweets and compiled a massive dataset with attachments, names, dates, replies, likes, retweets, etc. and released it to the public?
ignoring the legal side for a moment, if anyone edited or deleted something after the dataset was collected, then it would remain as it was before in the dataset. what if someone posted a private link, image, or other personal info on accident, the dataset caught that, no measures were put in place to censor or remove that data, and twitter refused to remove the data after the fact?
that's exactly what's happening here in this dataset.
because terminally online AI bros live in a bubble where they think ethics don't matter, the only thing that matters is owning the libs and training cute anime AIs.
They do already it just not public access you can pay for it though. If something was deleated after the fact than I should have been responsible enough to not post whatever it was, although deleting is subjective and most sites will still store your userdata (whatever was 'deleted').
It might be legal in the US, it's illegal in Europe in most cases, and users have a legal right to request (and obtain) their data to be removed from such datasets. Please check the GDPR laws. People here are simply demanding that these laws be respected, and you come tell them to swallow it and move on. Don't be that person.
Regardless of current laws, and how/where they apply, Open Source AI can and should encourage fairer practices than large closed-sourced tech companies.
Please read the HF ToS, Code of Conduct and Privacy Policy, you keep missing the mark on what's acceptable.
Not trying to come across that way, I said earlier the best form of action would be to contact the respective entities yourself. Im trying to provoke the question around data privacy itself, its not very good. IIRC in the US you can pretty much buy anyones info online. But regardless its still your responsability as a user to know how your data could be used by good or bad actors.
Not trying to come across that way, I said earlier the best form of action would be to contact the respective entities yourself. Im trying to provoke the question around data privacy itself, its not very good. IIRC in the US you can pretty much buy anyones info online. But regardless its still your responsability as a user to know how your data could be used by good or bad actors.
We know only too well. Still, EU citizens have extensive laws protecting us, so we can be online & expect companies to be liable for misusing our data. When we come across a violation of those rights, we request they are enforced.