#watermarking — Public Fediverse posts

This explainer shows how AI text watermarking works. commons.every.to/eb6c504586b3a7… #AI #Claude #watermarking

FACT: Anti-AI is bullshit and AI haters are simpletons.

“The people in favor of this watermarking for text have been sold a pipe dream, a fantasy. I’ve encountered dozens of comments from angry AI haters (many of them on Bluesky in particular, but also Threads and Hacker News) who are convinced that the only people who could be against the watermarking of AI-generated text are those who are duplicitously passing off AI-generated text as their own writing — and thus that I must be upset only because the jig will soon be up for me too. This of course is not true. I don’t even use AI to write text messages or emails for me, let alone a single sentence of my work.

But I find it funny that so many people who claim to believe that LLMs only produce “slop” and never anything useful also seem 100 percent convinced that the same LLMs are capable of watermarking their output in reliable ways. These people so desperately want to be able to point a finger at AI-generated text that they’ve fallen hook, line, and sinker for the argument from Google and Anthropic that, thanks to them, they’ll be able to.”

https://daringfireball.net/2026/08/follow-up_thoughts_on_watermarking

#AI #GenerativeAI #LLMs #Writing #EU #AIAct #Watermarking #Claude #Anthropic #Google

Lu l’entretien de Victor Storchan avec Scott Aaronson, "Pouvons-nous encore marquer les machines ?", dans Le Grand Continent. Quelque chose de plus proche de mon domaine pro.

https://legrandcontinent.eu/fr/2026/08/21/pouvons-nous-marquer-les-machines/

#Contrechat #NotesDeLecture #IA #Watermarking #AIAct

1/6

"… #Anthropic describes the act of #writing as being akin to a probabilistic game of chance. In Anthropic’s own words, its writing is sometimes the result of an “arbitrary random number generator,” and “random” whenever its systems encounter a situation where its tool believes, based on pattern recognition, that the choice between several possible next words isn’t all that important. That may be true for #LLMgarbage, but is not true for the human experience of writing, which is why human writing almost always feels different than #AIwriting."
#watermarking
https://www.404media.co/anthropics-text-watermarking-proves-ai-companies-do-not-care-at-all-about-writing/

"So we have a mistargeted law, badly drafted to go after a theoretical problem that hasn’t proven to be real, combined with penalties harsh enough that companies like Anthropic comply in the broadest, bluntest way possible — and all of it dropped into a world that has increasingly decided that AI use is a binary good-vs-evil question, and where no nuance is allowed.

The watermark won’t actually catch the people it’s ostensibly aimed at — those seeking to deceive people. Those people will likely be sophisticated enough to remove any such watermark and walk away clean. However, it will likely catch the people who had every right to use the tool in the first place: the non-native speaker who ran their draft through Claude to sound more fluent, the assistive-tech user who needed help composing a sentence, or the writer who wants an extra level of review on any text they’ve written. They’re the ones who will wear the scarlet letter. The EU wanted to look tough on AI and Anthropic wanted to look compliant. But neither of them has to answer to the the disabled user who used the tech to help them communicate, who now gets accused of lying and cheating."

https://www.techdirt.com/2026/08/20/the-eu-wanted-a-deepfake-detector-it-got-an-ai-scarlet-letter/

#AI #EU #AIAct #GenerativeAI #Watermarking #Deepfakes #NannyState #Anthropic #Claude #LLMs

Gizmodo: Anthropic Explains Its Watermark System as Some Claude Users Loudly Revolt. “In a blog post on Friday, Anthropic elaborated on its controversial watermarking system, announced earlier this week as an effort to comply with the European Union’s Artificial Intelligence Act. Watermarking will make it harder for users to get away with pretending they wrote AI-generated text.”

Anthropic's 'watermark' text adulteration in Claude is a perversion of writing (daringfireball.net)

https://daringfireball.net/2026/08/anthropics_watermark_text_adulteration_in_claude_is_a_perversion_of_writing

#writing #claude #anthropic #watermarking #eu

e564 with Andy, Michael and Michael - Stories and discussion on #Invisible #PromptInjection in #LegalBriefs & #resumes, #AI #watermarking, #PodcastGames, #LEGO and a whole lot more! #MinasTirith wasn’t built in a day!

http://gamesatwork.biz/2026/08/17/e565-building-minas-tirith/

e564 with Andy, Michael and Michael - Stories and discussion on #Invisible #PromptInjection in #LegalBriefs & #resumes, #AI #watermarking, #PodcastGames, #LEGO and a whole lot more! #MinasTirith wasn’t built in a day!

http://gamesatwork.biz/2026/08/17/e565-building-minas-tirith/

e564 with Andy, Michael and Michael - Stories and discussion on #Invisible #PromptInjection in #LegalBriefs & #resumes, #AI #watermarking, #PodcastGames, #LEGO and a whole lot more! #MinasTirith wasn’t built in a day!

e565 — Building Minas Tirith

e564 with Andy, Michael and Michael - Stories and discussion on #Invisible #PromptInjection in #LegalBriefs & #resumes, #AI #watermarking, #PodcastGames, #LEGO and a whole lot more! #MinasTirith wasn’t built in a day!

http://gamesatwork.biz/2026/08/17/e565-building-minas-tirith/

e564 with Andy, Michael and Michael - Stories and discussion on #Invisible #PromptInjection in #LegalBriefs & #resumes, #AI #watermarking, #PodcastGames, #LEGO and a whole lot more! #MinasTirith wasn’t built in a day! https://gamesatwork.biz/2026/08/17/e565-building-minas-tirith/

Jak odpowiedzi AI będą znakowane w sposób niewidzialny dla ludzi?

Niebawem wszystkie znane narzędzia AI zaczną dodawać w niewidoczne dla ludzi znaki do generowanych przez siebie odpowiedzi (zarówno tekstu, obrazów jak i kodu). Te tzw. znaki wodne (ang. “watermark”) to nie będą niewidzialne znaki Unicode, czy “mdashe” porozstawiane w odpowiednich miejscach ;) Anthropic, twórca Claude już opisał, jak zamierza to robić — i dlaczego musi to robić.
Claude zostawi w odpowiedziach ślady, ujawniające że z niego skorzystałeś na 2 sposoby
Anthropic potwierdził, że modele Claude wprowadzone w UE po 2 sierpnia 2026r. albo później będą obsługiwać “maszynowo odczytywalne oznaczenia”. Niewidoczny watermark trafi do całego tekstu odpowiedzi zarówno z aplikacji Claude Code/Cowork, jak również API i będzie też dodawany jeśli z modeli Anthropica ktoś będzie korzystał od innego dostawcy. Starsze modele są dopiero dostosowywane do tego wymogu, więc nie jest prawdą, że dziś “każda dodpowiedź Claude’a już jest znakowany”.
Powyższa ilustracja została syntetycznie wygenerowana z wykorzystaniem sieci neuronowej podczas procesu inferencji modelu generatywnego typu text-to-image
Anthropic wdraża dwa mechanizmy.

Generowany tekst otrzymuje na poziomie modelu znak, który przechodzi przez kopiuj-wklej i może przetrwać część edycji.
Generowana grafika (PNG, JPG i SVG) dostaje natomiast podpisane metadane pochodzenia C2PA. Podpis informuje, że plik {prze,s}tworzył Claude, ale może zniknąć po konwersji, ponownym zapisaniu lub wykonaniu zrzutu ekranu.

Nie szukaj niewidzialnych znaków — to na nic!
Anthropic nie ujawnił szczegółów technicznego mechanizmu watermarku. Nie ma więc podstaw, by twierdzić, że są to “niewidoczne znaki Unicode” usuwalne prostym parserem. [...]

#AI #Anthropic #ChatGPT #Claude #Fingerprinting #Gemini #Google #OpenAI #Watermarking

https://niebezpiecznik.pl/post/jak-odpowiedzi-ai-beda-znakowane-w-sposob-niewidzialny-dla-ludzi/

New article: Digital Signatures and Watermarks — Defending Authenticity in the Era of Generative AI.

As generative AI makes convincing fakes easier, robust authenticity tools become essential. This piece explains how digital signatures and watermarking work, where they fit in compliance and risk strategies, and what organizations should consider when implementing them. Read the full article: https://wix.to/XM8P8vJ

#AI
#Watermarking
#ContentIntegrity
#DigitalSignatures
#RiskManagement
#InformationSecurity

#shownotes for @gamesatwork_biz #podcast e565 with are done and publication set for tomorrow on https://www.gamesatwork.biz together with @michaelrowe01 and @andypiper. Find e565 on @Spotify @overcastfm @YouTube and all your favorite podcast feeds! Stories and discussion on #Invisible #PromptInjection in #LegalBriefs & #resumes, #AI #watermarking, #PodcastGames, #LEGO and a whole lot more! #MinasTirith wasn’t built in a day! Subscribe on https://www.gamesatwork.biz so you don’t miss an episode!

Pulling together the #shownotes in #ChapelHill for Monday’s posting of @gamesatwork_biz episode e565 with @andypiper, @michaelrowe01 and yours truly. Stories and discussion on #Invisible #PromptInjection in #LegalBriefs & #resumes, #AI #watermarking, #PodcastGames, #LEGO and a whole lot more! #MinasTirith wasn’t built in a day! Check out earlier episodes, chock full of #AI #metaverse #AR #VR #gamification and so much more on https://www.gamesatwork.biz

AI Text Watermarking Is Not a Big Deal, by @seangoedecke.bsky.social:

https://www.seangoedecke.com/ai-text-watermarking-is-not-a-big-deal/?ref=frontenddogma.com

#ai #watermarking #anthropic

#AI #text #watermarking works by embedding a pattern in the choices between words made by a language model. A secret key subtly influences these choices, creating a detectable pattern without altering the text’s meaning. This #watermark can be verified by the key-holder, but is erased by significant #editing or #recomposition. https://declaude.org/watermarking/?eicker.news #tech #news #ainews

This is a cool explainer, and it’s interactive, too!

#ai #watermarking

https://declaude.org/watermarking/