upvote
Doesn't help if someone you sent information to pastes it into the chatbot under their not-opted-out account.
reply
Well if you have some specific confidentiality requirements for some information, don't send it to people who you aren't 100% sure will respect those requirements.

And if you're publishing something in general, someone feeding it to a chat bot is really low on the scale of things to be concerned about.

reply
What if someone has mail connected and it reads fhe emails this nonsensical you cant share even bits of i formation now huh?

Same issue with phones they transcibe and allow summarizing calls who knows which models they use.

reply
That's peak of AI psychosis, you know.

And if anyone actually sold such phones, I'd be first to buy it. LLMs are godsend for dealing with daily communications, but this process is still mostly manual on end-user devices.

reply
Professional mathematicians just don't believe this.
reply
And what's your interpretation? Ie. if they convert/summarize your data and train on that – is it "your" data or "theirs" now? What is their interpretation?
reply
Does it matter? Only matters if a court will ever fine them enough to care, which they wont.
reply
Why would anybody possibly believe this?
reply
Anthropic and OpenAI stole all the information from the Internet yesterday, but this is today.
reply
Except Fable
reply
can someone flag this bs please.
reply

    [ ] Check this box and we'll *totally* not use your data for training
They might as well have this box as well

    [ ] Check this box if you'd like to preemptively be paid $25 as the result of a future class-action lawsuit because we just used the data anyway
reply
$25 would be a record for me. $5, tops.
reply
$25 worth of fable tokens!
reply
Even though I'm not as cynical as you on this, this is S-tier humor. All privacy choices everywhere should be framed this way. I wonder if I framed all of the notices of data breaches and slap-on-the-wrist settlement privacy violation notices I've ever gotten, if I could cover the walls of a whole room with the frames at this point.
reply
Cynical or realistic? Major tech companies have been caught doing that hundreds of times.
reply
So you trust the biggest information hoovers of our time to always respect such settings?
reply
Yes, because they need enterprise money more than anything.
reply
Enterprise accounts are a whole different ballgame though. Not at all implausible they’d do one and not the other.
reply
It still would be toxic to corporate decision makers if they were caught dishonoring this control on one type of an account. If they're willing to lie and cheat on one tier of account, what guarantees they won't lie and cheat on the other tier next quarter. Especially with Anthropic being a public company now, that pressure would certainly be there if their corporate culture were comfortable with this type of fraud.
reply
But didn’t they already train their AI systems by violating copyright, vacuuming up any and all material they could find? Hasn’t that bridge already been crossed?
reply
There's a difference between "These guys probably downloaded a few hundred thousand books on BitTorrent in order to get this smart" and "These guys will train on my company's internal communications and thus our knowledge and wisdom and trade secrets will be made available to others."

Tangent: I personally literally don't care that all these LLMs trained on a ton of copyrighted material. It is in my humble opinion a long-awaited dose of karma for the copyright lobby, whose own abuses over the decades are numerous. It has basically restored balance to a playing field where the public has had no rights and (mostly) publishing conglomerates have held all the rights and privileges.

reply
No, except in the minds of people who suddenly became copyright maximalist out of envy, and adopted a "dog in the manger" attitude.
reply
Such privacy checkboxes always seem to suffer from a string of bugs that causes them to reset to the "suck all my data" option. It's really quite strange.
reply
I think they need to prove relevancy and potential for future disruption more than anything to keep IPOs and investments big.

That means stealing leading researchers info is probably great for them, whether it's intentional or not.

reply
Also they can adhere to the settings of no training and still use your data otherwise.

They have all the best ideas flowing through there servers in real time as they are hatched around the world. It would be tempting to have top scientists accounts posted to internal slack channels to monitor new discoveries.

reply
Yes, also why is that even a topic here?

Don't post confidential information where there's a risk of confidentiality being broken. That includes posting to people whom you aren't 100% sure of know and uphold your confidentiality requirements. Adults usually handle this by contracts with penalties that include financial damages, loss of employment, or prison time.

reply
Quick public service announcement: both have shown that they could not care less what we turn off, or even for the legal slaps on the wrist they'll get for saying "oops, turns out we did read them".
reply
Encrypting your email might make intelligence agencies look at you that much harder.
reply
OK, then they just have a smaller LLM summarize everything you send them and train off the derivative.

see? easy peasy for them to adhere to "no training".

I wouldn't doubt OpenAI and Anthropic, etc all have conversation or message categorization like they do for flagging bio, weapons, content but use it internally to flag ground breaking maths in real time and kick the convo up the chain of command for further review.

reply
Call me crazy, but I kinda don't really think they need it.

People talk as though all "data" is valuable to AI companies, but ingesting arbitrary bytes into LLM training today, en masse seems like a recipe for degrading its quality.

Even in the domain of mathematics papers. When the frontier models are already really great at this maths stuff, feeding in every blob of text someone grammar-checks or whatever doesn't improve it. You don't know up front if the text is from some crackpot, from a naïve student with minimal understanding, or from the greatest mind of the generation. Though the chances are good that it's not that last one.

It seems like you'd be interested in ingesting published papers that are either from highly respected authors, or from newer authors which the eminent minds have shown respect and praise for. But I just don't buy that if someone pasted in some text, or if Claude Desktop read a draft of this paper in someone's email or whatever, that we'd find that a model had been trained on it.

reply
I have a bridge to sell you
reply