Emails are also instructions to a computer-based service (SMTP) that you presumably signed your rights away to when you accepted the T&Cs.
Yet no one would think it's acceptable for the NYT and a dozen other news organizations to request an "anonymized" archive of all your emails from provider X, just because said provider is in a lawsuit with them, and you have nothing to do with any of it.
This is shameful, and would create a dangerous precedent. Really hope the order gets struck down.
Well yes, that sort of evidence is routinely used to gather evidence and build criminal cases. Emails, like letters, are correspondence between individuals.
ChatGPT isn't (despite it's name) equivalent - the nearest analogy is Google. We know the modus operandi of the world based on these services (incl social media) and privacy is the aspect that's been given up.
The linked article is making the point that the judge is claiming that they are preserving privacy with this order because they believe in de-anonymization of the data set. The judge appears to have no understanding of how re-identification works and the history here.
So while it's an interesting question about whether privacy exists, the point here is that it doesn't exist, but the judge is saying it does.
In the narrative about this the affordance given to OpenAI is staggering and the naivety that "stuff on the web will stay private" is contra to everything we've learnt over the past 25 years though.
never say anything to a cloud hosted chatbot you wouldn't want to see printed in the new york times
anonymization wouldn't remove personally identifying data typed in. I can think of lots of examples.
If it ain't zero knowledge, it's a corporate surveillance state bonanza.
If it's free or too cheap from a corporation, it's too expensive.
Wouldn't every single American in that 20M people have standing (legal term) to prevent this?
Related:
Fighting the New York Times' invasion of user privacy
https://news.ycombinator.com/item?id=45900370
We analyzed 47,000 ChatGPT conversations. Here's what people use it for
https://news.ycombinator.com/item?id=45902767
In what way are these "private chats"?
These are instructions to a computer-based service that you presumably signed your rights away to when you accepted the T&C's?
Emails are also instructions to a computer-based service (SMTP) that you presumably signed your rights away to when you accepted the T&Cs.
Yet no one would think it's acceptable for the NYT and a dozen other news organizations to request an "anonymized" archive of all your emails from provider X, just because said provider is in a lawsuit with them, and you have nothing to do with any of it.
This is shameful, and would create a dangerous precedent. Really hope the order gets struck down.
Well yes, that sort of evidence is routinely used to gather evidence and build criminal cases. Emails, like letters, are correspondence between individuals.
ChatGPT isn't (despite it's name) equivalent - the nearest analogy is Google. We know the modus operandi of the world based on these services (incl social media) and privacy is the aspect that's been given up.
The linked article is making the point that the judge is claiming that they are preserving privacy with this order because they believe in de-anonymization of the data set. The judge appears to have no understanding of how re-identification works and the history here.
So while it's an interesting question about whether privacy exists, the point here is that it doesn't exist, but the judge is saying it does.
They were chats that were supposed to be private between you and OpenAI, concretely. Nobody, including OpenAI, expected them to ever become public.
In the narrative about this the affordance given to OpenAI is staggering and the naivety that "stuff on the web will stay private" is contra to everything we've learnt over the past 25 years though.