Guardian and Verge probe whether the watermark hurts quality
8 Aug 17 12:53 PM · 62d ago · 15 articles · 61 posts · 55 comments · 6 sources · development 8 of 8
The Guardian and The Verge published explainers on the mechanics of the watermark and questioned whether the technique makes AI writing worse, while some commentators pushed back that the outrage over being 'caught' was overblown.
“every needs to go outside, breath, and touch some grass for a few minutes”
@ianriccaboni.comAnthropic AI lab, maker of ClaudeJohn Gruber Daring Fireball writerArturo Villarroya Claude Max subscriber, public policy workerVladislav Rajtmajer Freelance developer, Czech RepublicJames Padolsey Author of interactive watermarking explainer
The whole story articlespostscomments the bright band is this development · numbered dots are the others · click one to jump
Reported in the same hours no headline names this development itself — these 5 claims were published in its stretch
-
first by NPR, 62d ago · also Android Authority, The Verge, Mashable
4 more headlines
- Anthropic reveals how Claude secretly watermarks AI-written text Android Authority · 62d ago
- Anthropic explains how Claude’s invisible text watermarks will work The Verge · 62d ago
- What Claudes AI text watermark actually does Mashable · 62d ago
- What Claude's AI text watermark actually does Mashable · 62d ago
-
first by The Guardian, 62d ago · also Guardian Media, Guardian Tech
-
first by inc.com, 62d ago · also The Decoder
1 more headline
- Anthropic watermarks Claude's output, but critics question the tradeoffs The Decoder · 62d ago
-
first by Business Insider, 62d ago
-
first by The Hindu, 62d ago
What people said 24 voices · best of 73 · verbatim
-
A
can't quite put my finger on why but this has gotta be the funniest, most absurd, most humiliating sentence i've read in a long long time (from https:// daringfireball.net/2026/08/ant hropics_watermark_text_adulteration_in_claude_is_a_perversion_of_writing )
-
I agree with the substance of this article, and disagree with the author's reaction to it.The part I agree with:It is true that watermarking can be done by "just" swapping one PRNG for another, and it is even true that with today's LLMs, it is possible that this will not degrade the output. But it has a cost, and as things improve, that cost will…
-
J
"a paper in a leading chemistry journal had to be retracted because the authors appeared to have used an AI tool to draft part of the publication. The tool used the phrase “mass killing of an ethnic group” as an alternative for “final solution”, though the paragraph in question appeared to be about a zinc nanogel." # AI https://www…
-
I commented it last time the post about Claude watermarking went viral and I'm going to say the same thing again:"I've been working in the media and model IP space for quite many years. What this article misses a bit is the threat model for watermarking in general. Watermarking and fingerprinting have inherently weak security guarantees -- they…
-
S
Aus dem Feedreader: ★ Anthropic’s ‘Watermark’ Text Adulteration in Claude Is a Perversion of Writing https:// daringfireball.net/2026/08/ant hropics_watermark_text_adulteration_in_claude_is_a_perversion_of_writing # bookmark # rss
-
> And even that is not sufficient, as the text might have been generated with ChatGPT, Gemini, Grok, MistraLI would think that these things would eventually converge and we’d get one watermarking algorithm as an industry standard. That way, all major provider would follow it and we’d get independent software for checking. This would partly limit…
-
R
"This isn’t just about text one might generate with the intention of passing it off as their own natural work. This isn’t even about LLM proofreading of work written by hand. Anthropic is saying that all new Claude models are going to adulterate every single bit of text longer than 200 tokens (~150 words) they generate, including everything it…
-
I’m surprised by the comments here being so favorable to anthropic. The comments are right about there being no “best” token, and yeah Gruber may have an agenda here. But I think the fundamental principle is that this approach messes with the distribution in ways that deviate from the trained model. Take the “gray” and “overcast” choices. And lets…
-
M
Posts updated today: How Claude Marks AI-Generated Content https:// mjtsai.com/blog/2026/08/11/how -claude-marks-ai-generated-content/#how-claude-marks-ai-generated-content-update-2026-08-17 Touch Bar https:// mjtsai.com/blog/2026/05/05/the -problem-with-the-touch-bar/#the-problem-with-the-touch-bar-update-2026-08-17 Proof of Erdős Unit Distance…
-
> “By definition it must make text worse … because the nature of the watermarking algorithm requires it to sometimes increase the probability of selecting a worse word choice and decrease the probability of selecting the model’s best choice.”Gruber made an effort to but doesn't fully understand how SynthID works. LLMs select the next word randomly…
-
J
how about Captain Jack Kane what's what's the record authority did six but I think you can do ten and starting right now sorry about that all right now an OB license and registration your registration hurry up now sorry no well then why you laughing mr. Larry Johnson all right now where were we I'm sorry you see yeah I see now I thought you don't…
-
> I want any LLM I use to choose the very best, most precise words at every single decision point.Oh! If you want that, you should run your own model and set the generation temperature to 0 :) Because that's not what any commercial LLM is doing. Never has been. This is just making up a universe that doesn't exist so you can get mad about no longer…
-
F
I understand the sentiment of https:// mastodon.social/@daringfirebal l@mastodon.social critique of Anthropic's watermarking. The choice of words it generates is influenced by the method to create a watermark. In other words, it's the best it can do, minus the requirements of the watermark. But the problem is, this assumes that "the best it can…
-
> “He leaped at the chance” and “He jumped at the opportunity” are very similar sentences expressing the same general sentiment, but they are not the same. The exact words we choose when writing matter. I want any LLM I use to choose the very best, most precise words at every single decision point.Neither of these is "better" or "more precise"; in…
-
M
It's hard to imagine why someone would be angry at # AI watermarking. This technology is essential to combating AI slop and also to the companies themselves to prevent model collapse. And yet, here is John Gruber with yet another knee-jerking anti-EU piece. AI text watermarking is just making some words a bit more probable when outputted by…
-
Why not switch it around? It's far more important to be able to verify content has not been tampered with after originating from a human or capture device (like a camera) and we have decades of cryptographic experience in reliably signing text, images, video, etc.Having proof that imagery is unmodified (whether via Photoshop, Paint or a model) is…
-
C
My god. https:// daringfireball.net/2026/08/ant hropics_watermark_text_adulteration_in_claude_is_a_perversion_of_writing "The exact words we choose when writing matter. I want any LLM I use to choose the very best, most precise words at every single decision point." Bruh, what!? You're literally not choosing your own words when you use an LLM. You…
-
My biggest concern is that checking any text for watermarks requires sending the entire text to Anthropic. And even that is not sufficient, as the text might have been generated with ChatGPT, Gemini, Grok, Mistral, ...So every check requires sending the text to as many AI providers as offer a watermarking detection API, almost all of which have a…
-
T
So you have “concerns that content generated or revised with Claude could still be identified as AI-assisted even after the text is reworked, reformatted, or incorporated into another piece of writing”? THEN STOP USING THE CRAPPY PLAGIARISM TECH THEN. Learn to create. Learn to express yourself without it. You don’t need it, it needs you - blindly…
-
One could also use butterflies to write ;) https://xkcd.com/378/The problem is, that LLMs worked very well for me to improve my writing. Especially as I'm not a native speaker, it was a great way to improve the legibility of my work.I want a tool that helps me improve my writing. A tool I can learn from. Not a tool that switches out "bananas" to…
-
'The exact words we choose when writing matter. I want any LLM I use to choose the very best, most precise words at every single decision point" - then, err, write it yourself? (It's especially weird to see an Apple cheerleader weeping about a company overriding user autonomy.) https:// daringfireball.net/2026/08/ant…
-
What a bunch of entitled whining. How is the system to know that it's just a private conversation that won't be used in some fraudulent way? Abuse is currently rampant, yes please let's find a way to mark LLM output. The thing I'm worried about is giving the providers the power to claim provenance. Even ignoring the privacy issues, the operational…
-
S
TL;DR: Anthropic reveals its implementation of invisible text watermarks in Claude for compliance with Europe's AI transparency regulations, utilizing an approach based on Google's SynthID-Text technology. https://www. theverge.com/ai-artificial-int elligence/980869/anthropic-claude-watermarks-synthid-text-system # law # tech # legaltech ⚖️ 🤖 #…
-
The best discussion I've seen so far is from Scott Aaronson: https://scottaaronson.blog/?p=6823> To illustrate, in the special case that GPT had a bunch of possible tokens that it judged equally probable, you could simply choose whichever token maximized g [a cryptographic function]. The choice would look uniformly random to someone who didn’t…
All 8 developments of Claude's AI text watermark sparks backlash, cancellations… →
Hacker NewsNewswiresMastodonRedditXBlueskyGoogle NewsLobsters