Started By
Message

re: Claude turning snitch?

Posted on 11/1/26 at 10:41 am to
Posted by SlowFlowPro
With populists, expect populism
Member since Jan 2004
487635 posts
Posted on 11/1/26 at 10:41 am to
quote:

Does this mean that students using AI to do their assignments can be detected easier?

Yes.
Posted by Joshjrn
Baton Rouge
Member since Dec 2008
34164 posts
Posted on 11/1/26 at 10:41 am to
quote:

you don't think that digital signature could be traced back to you and then they decide your politics or requests don't meet "community standards" then the next thing you know you've been debanked and turned down for ride shares by Uber. But keep that head in the sand. This is just used to keep college kids from cheating on their essays.

It literally says in the OP that there is no user identifying information. That could change, or it could be a lie, but there’s nothing in the OP I’m particularly concerned about. Frankly, the fact that they are announcing this publicly is going to reduce demand for their product.
Posted by Roaad
White Privilege Broker
Member since Aug 2006
86253 posts
Posted on 11/1/26 at 10:56 am to
Good
Posted by UnitedFruitCompany
Bay Area
Member since Nov 2018
4427 posts
Posted on 11/1/26 at 1:49 pm to
I know a guy. Owns a company. He used chatgpt to write his employee handbook. The a-hole legit copy pasted word for word what chat gpt wrote. Even the part where it said, "take this section out. Only for management". Also had a section in there about not using AI for personal info. What did he tell all his employees to do? Sync their email and cloud storage, which contained all their customer info, with chatgpt.

He's now getting sued into oblivion by an employee he fired who knew all of this, pointed it out, and it pissed the owner off so he fired the guy.


To me this a good thing. frick these guys that think AI is some fairy godmother that can solve all your problems in a chat window.
Posted by Centinel
Idaho
Member since Sep 2016
47378 posts
Posted on 11/1/26 at 2:31 pm to
quote:

that text will be identifiable as generated by Claude if you run it through an algorithm that Anthropic is providing to certain entities


This is the problem I have. "Certain entities."
Posted by GetMeOutOfHere
Member since Aug 2018
1198 posts
Posted on 11/1/26 at 2:32 pm to
quote:

When Claude generates text, it’s going to do so in a way that the combination of characters in that text will be identifiable as generated by Claude if you run it through an algorithm that Anthropic is providing to certain entities for the purpose of determining whether a given text is AI generated.


You're not wrong to worry about AI watermarks -- it is a serious concern.

It's not just an AI detection, it's checking about the sincerity of the person writing the message.

I would say you are absolutely right.
Posted by Volvagia
Fort Worth
Member since Mar 2006
53908 posts
Posted on 11/1/26 at 2:36 pm to
quote:

Not this system


Sure you can. You just have to paraphrase the product in your own words. It’s not like a hidden digital tag. It’ll be like tracking patterns of how many syllables and letters words have and their frequency. The algorithm will adjust its output to meet those parameters.
Posted by skullhawk
My house
Member since Nov 2007
28360 posts
Posted on 11/1/26 at 2:45 pm to
This seems like it will take a bite out of their business.

I get where they are going, ownership of everything it produces, but a lot of people are going to switch to a competitor.
Posted by Rhio
Lake Charles
Member since Dec 2013
1443 posts
Posted on 11/1/26 at 5:13 pm to
I honestly thought AI already did this. I’m surprised it took this long.
Posted by FLTech
he/won
Member since Sep 2017
30166 posts
Posted on 11/1/26 at 5:44 pm to
So let me get this straight.. people keep saying that AI is going to end the world but then when they put up safeguards, you people bitch that we are so fricked

Make up your fricking minds
Posted by Bestbank Tiger
Premium Member
Member since Jan 2005
84006 posts
Posted on 11/1/26 at 5:46 pm to
How do we give it stitches?
Posted by wheelr
New Iberia
Member since Jul 2012
6293 posts
Posted on 11/1/26 at 5:55 pm to
quote:

You're not wrong to worry about AI watermarks -- it is a serious concern.

It's not just an AI detection, it's checking about the sincerity of the person writing the message.

I would say you are absolutely right.


He is justified to push back.

It's not paranoia. It's the right thing to do.
Posted by CAD703X
Liberty Island
Member since Jul 2008
95310 posts
Posted on 11/2/26 at 8:52 am to
quote:

literally says in the OP that there is no user identifying information.
Posted by CAD703X
Liberty Island
Member since Jul 2008
95310 posts
Posted on 11/2/26 at 8:57 am to
quote:

Holy
fricking
shite
lawfare doesn't exist.

Nobody was debanked for supporting Trump in 2020

Parents weren't put on FBI watch lists for showing up at school board meetings to protest perverts using girls bathrooms.

January 6 was worse than 911 and Pearl Harbor combined.

Enjoy your AI UUIDs tracking everything you've ever had their help on, corporations would never share that info to help governments track your digital footprint.
This post was edited on 11/2/26 at 9:05 am
Posted by Joshjrn
Baton Rouge
Member since Dec 2008
34164 posts
Posted on 11/2/26 at 9:49 am to
quote:

CAD703X

I’m not exactly one to buy into corporate bullshite, but it does elicit the question of why they would offer this information sua sponte and then lie about how the algorithm works. They could have just, you know, not told you
Posted by Poboy1911
Member since Aug 2026
146 posts
Posted on 11/2/26 at 10:28 am to
lol, people will have a way to remove this watermark within a week of rollout
Posted by Stinger_1066
On a golf course
Member since Jul 2021
4323 posts
Posted on 11/2/26 at 11:26 am to
quote:

How is AI outing its work product as AI an example of AI imposing its will on us?


Just my educated guess, but this is what I think:

1 - they will use this to try to expose competitors that are stealing their product.

2 - if you use their product and are monetizing your work, they can claim a portion of the profits if you don't have a specific type of licensing arrangement with them.

Think of it this way: suppose you write a book that you sell on-line in Adobe Acrobat format, and Adobe came along and sued you for a percentage of the profit you make on the book.
Posted by mattz1122
Member since Oct 2007
57349 posts
Posted on 11/2/26 at 11:43 am to
Wroten? Never.
Posted by Bamafig
Member since Nov 2018
7280 posts
Posted on 11/2/26 at 3:37 pm to
Does this mean kids will have to do their own homework?
Posted by lsuconnman
Baton rouge
Member since Feb 2007
5934 posts
Posted on 11/2/26 at 3:51 pm to
It will be just the opposite. As already stated, you can’t copyright AI created work.

Practically all original thought has already been scraped from the web. This will allow Claude to skip AI generated content it already created and focus on stealing whatever original thought is left. It will probably also be able to filter competitors AI generated information to reverse engineer it.
first pageprev pagePage 2 of 3Next pagelast page

Back to top
logoFollow TigerDroppings for LSU Football News
Follow us on X, Facebook and Instagram to get the latest updates on LSU Football and Recruiting.

Facebook•X•Instagram