\
  The most prestigious law school admissions discussion board in the world.
BackRefresh Options Favorite

BREAKING: Anthropic shuts down consuela's substack (link)

https://support.claude.com/en/articles/16266773-how-claude-m...
Nighttime weed whacker
  08/10/26
...
sepia exhilarant dopamine
  08/10/26
...
Confused senate
  08/10/26
...
swashbuckling field jap
  08/10/26
LOL @ the EU. OI M8 U GOT A LOISSENCE FOR THAT AI SLOP?
high-end friendly grandma locale
  08/10/26
i actually think this is pretty cr it's not a perfect &qu...
Nighttime weed whacker
  08/10/26
...
Frum Hospital
  08/10/26
Seems like the lib instinct to "do something" even...
high-end friendly grandma locale
  08/10/26
lol
Nudist personal credit line dragon
  08/10/26
lmao @ anyone using American AI, time to switch to Chinese L...
Vibrant Abusive Kitchen
  08/10/26
I don't think open ai is retarded enough to do this
Nudist personal credit line dragon
  08/10/26
...
Confused senate
  08/10/26
they will follow suit imo. they'll have to
Nighttime weed whacker
  08/10/26
Open AI has researched this exact kind of statistical token ...
Nudist personal credit line dragon
  08/10/26
those "drawbacks" in question? being at a marke...
Nighttime weed whacker
  08/10/26
You know how it says in this article it doesn't rely on meta...
Nudist personal credit line dragon
  08/10/26
yeah there are plenty of issues with it it was still a fu...
Nighttime weed whacker
  08/10/26
...
Confused senate
  08/10/26
...
Confused senate
  08/10/26


Poast new message in this thread



Reply Favorite

Date: August 10th, 2026 4:09 PM
Author: Nighttime weed whacker

https://support.claude.com/en/articles/16266773-how-claude-marks-ai-generated-content

Anthropic says new Claude models will embed invisible watermarks in all generated text, everywhere Claude is offered.

The watermark is part of the text, it isn't metadata: "it will travel with the text when it's copied and pasted elsewhere, and may persist through some editing."

This starts with models launched on or after August 2, 2026, under an EU AI Act code Anthropic signed. Anthropic is still working on adding it to current models. The rollout is worldwide.

(http://www.autoadmit.com/thread.php?thread_id=5891475&forum_id=2phpsessid=49d358651c#50058415)



Reply Favorite

Date: August 10th, 2026 4:11 PM
Author: sepia exhilarant dopamine



(http://www.autoadmit.com/thread.php?thread_id=5891475&forum_id=2phpsessid=49d358651c#50058419)



Reply Favorite

Date: August 10th, 2026 4:12 PM
Author: Confused senate



(http://www.autoadmit.com/thread.php?thread_id=5891475&forum_id=2phpsessid=49d358651c#50058422)



Reply Favorite

Date: August 10th, 2026 4:14 PM
Author: swashbuckling field jap



(http://www.autoadmit.com/thread.php?thread_id=5891475&forum_id=2phpsessid=49d358651c#50058426)



Reply Favorite

Date: August 10th, 2026 4:16 PM
Author: high-end friendly grandma locale

LOL @ the EU.

OI M8 U GOT A LOISSENCE FOR THAT AI SLOP?

(http://www.autoadmit.com/thread.php?thread_id=5891475&forum_id=2phpsessid=49d358651c#50058429)



Reply Favorite

Date: August 10th, 2026 4:18 PM
Author: Nighttime weed whacker

i actually think this is pretty cr

it's not a perfect "solution" by any means but it will be a net improvement compared to not having it

(http://www.autoadmit.com/thread.php?thread_id=5891475&forum_id=2phpsessid=49d358651c#50058433)



Reply Favorite

Date: August 10th, 2026 4:19 PM
Author: Frum Hospital



(http://www.autoadmit.com/thread.php?thread_id=5891475&forum_id=2phpsessid=49d358651c#50058438)



Reply Favorite

Date: August 10th, 2026 4:24 PM
Author: high-end friendly grandma locale

Seems like the lib instinct to "do something" even if it's just performative. All people need to do is figure out how the watermark is embedded into the text and then change a couple words. Meanwhile, over time these "can't hurt, why not?" type laws increase compliance costs makes everything more expensive and harder to compete internationally.

(http://www.autoadmit.com/thread.php?thread_id=5891475&forum_id=2phpsessid=49d358651c#50058460)



Reply Favorite

Date: August 10th, 2026 4:21 PM
Author: Nudist personal credit line dragon

lol

(http://www.autoadmit.com/thread.php?thread_id=5891475&forum_id=2phpsessid=49d358651c#50058445)



Reply Favorite

Date: August 10th, 2026 4:19 PM
Author: Vibrant Abusive Kitchen

lmao @ anyone using American AI, time to switch to Chinese LLMs

(http://www.autoadmit.com/thread.php?thread_id=5891475&forum_id=2phpsessid=49d358651c#50058439)



Reply Favorite

Date: August 10th, 2026 4:21 PM
Author: Nudist personal credit line dragon

I don't think open ai is retarded enough to do this

(http://www.autoadmit.com/thread.php?thread_id=5891475&forum_id=2phpsessid=49d358651c#50058444)



Reply Favorite

Date: August 10th, 2026 4:22 PM
Author: Confused senate



(http://www.autoadmit.com/thread.php?thread_id=5891475&forum_id=2phpsessid=49d358651c#50058449)



Reply Favorite

Date: August 10th, 2026 5:37 PM
Author: Nighttime weed whacker

they will follow suit imo. they'll have to

(http://www.autoadmit.com/thread.php?thread_id=5891475&forum_id=2phpsessid=49d358651c#50058708)



Reply Favorite

Date: August 10th, 2026 4:20 PM
Author: Nudist personal credit line dragon

Open AI has researched this exact kind of statistical token based "watermarking" and decided not to use it because there are too many major drawbacks

(http://www.autoadmit.com/thread.php?thread_id=5891475&forum_id=2phpsessid=49d358651c#50058441)



Reply Favorite

Date: August 10th, 2026 4:21 PM
Author: Nighttime weed whacker

those "drawbacks" in question?

being at a market disadvantage compared to your competitors who don't implement the watermark

(http://www.autoadmit.com/thread.php?thread_id=5891475&forum_id=2phpsessid=49d358651c#50058446)



Reply Favorite

Date: August 10th, 2026 4:27 PM
Author: Nudist personal credit line dragon

You know how it says in this article it doesn't rely on metadata? It's because the way that it works is they put in an algorithmic bias toward selecting certain tokens at certain spots that supposedly don't change the meaning of the output, and that makes it traceable as coming from that model. But if you rewrite parts of it or put it through a translator and back it destroys it. Another thing is they were worried it would have a DITHPARTE IMPACT on ESL speakers who can't desloppify AI outputs well. I'm sure they also considered that people won't want to use it anymore if they know its easily traced from claude.

(http://www.autoadmit.com/thread.php?thread_id=5891475&forum_id=2phpsessid=49d358651c#50058474)



Reply Favorite

Date: August 10th, 2026 4:30 PM
Author: Nighttime weed whacker

yeah there are plenty of issues with it

it was still a funny joke and also just a fact that the competitive disadvantage is by far the most important consideration in play here

(http://www.autoadmit.com/thread.php?thread_id=5891475&forum_id=2phpsessid=49d358651c#50058484)



Reply Favorite

Date: August 10th, 2026 4:32 PM
Author: Confused senate



(http://www.autoadmit.com/thread.php?thread_id=5891475&forum_id=2phpsessid=49d358651c#50058489)



Reply Favorite

Date: August 10th, 2026 4:22 PM
Author: Confused senate



(http://www.autoadmit.com/thread.php?thread_id=5891475&forum_id=2phpsessid=49d358651c#50058451)