Claudeの電子透かしとは？仕組みと検出でわかることの限界

"Aug 14, 2026"（公開日）／"When watermarking is used, choices are still made at random, but the source of the randomness is different. Instead of using an arbitrary random number generator to pick the next word, watermarking uses the key and a few words that come before to settle what word the model should pick."（What is watermarking?）／"Importantly, it isn't that the model will now always be biased toward overcast or grey."（同節）／"Nothing is added to the text and there are no hidden characters;"（To summarize）／"Claude's text watermark is a version of the SynthID-Text approach published by Google DeepMind in a Nature paper in 2024. It belongs to a family of approaches that go back to a proposal by Scott Aaronson in 2022, all of which share the same design principle that we described above—the watermark only changes the source of the randomness used to pick among words."（Which specific method of watermarking do you use?）— Anthropic 公式発表より

"To a reader, a watermarked response is indistinguishable from an unwatermarked one"／"In internal testing, we've seen no impact of watermarking on the content, level of creativity, or readability of Claude's text. In the SynthID-Text paper, which introduced the technique we use, Google DeepMind tested this impact by serving a model that used watermarking to a portion of their Gemini traffic and comparing thumbs-up and thumbs-down ratings. They found no statistically significant differences from the unwatermarked model. And in a controlled study, human raters comparing watermarked and unwatermarked answers side-by-side saw no difference in quality."（How does watermarking affect Claude's outputs?）／"Watermarking has a negligible impact on the speed of models, and because it produces no extra tokens, the model is the same price to serve and use."（Does this slow the model down, or make it more expensive?）— Anthropic 公式発表より

"The EU law includes a transition period for Anthropic models launched before August 2, 2026, and we're working to add watermarking for those models as well. This will be rolled out over the coming months."（What about older Claude models?）／"We're applying watermarking globally at launch because we don't yet have a durable way to scope it by region."（Why are you watermarking Claude's outputs?）— Anthropic 公式発表より

"Products. Claude markings cover output from supported models everywhere you use Claude, including Claude Platform (API), Claude, Claude Code, Claude Cowork, and Claude Tag. Embedded watermarks will apply to all generated text."／"Cloud partners. Embedded watermarks will apply when supported Claude models are accessed through AWS, Google Cloud, or Microsoft Foundry."／"Watermarking will be applied at the model level, which means it will be present no matter which Claude product or surface the text comes from."（What's covered ／ 1. Embedded watermarks in text）— Anthropic サポート文書より

"Watermarking is sparser on factual passages where there are fewer choices that can be made without decreasing the accuracy of the text."／"so the watermark would have nothing to act on."（いずれも Which specific method of watermarking do you use?）／"once the model has written "2 + 2 =", there is a very clear best choice for the next token"／"The "nudge" of the watermark wouldn't be applied here."（いずれも What about code?）— Anthropic 公式発表より

"code—which in very many cases has to be exact—has generally less watermarking than some other forms of text. Having said that, in areas where there is an arbitrary choice between particular words or terms within the code, the watermark can be used, such as comments within code."／"But by definition, it will have a negligible effect on the actual code produced."（いずれも What about code?）— Anthropic 公式発表より

"The watermark only applies to words Claude chooses. When Claude proofreads text written by a person, what it gives back has generally only been lightly edited; because nearly all the words are the person's, there's very little (if anything) for the watermark to attach to."／"Depending on the length of the text and how heavily Claude has edited it, those changes might not be enough to make Claude's involvement detectable. The more Claude writes, the more decisions it has to make, and the more space there is for a watermark."（いずれも What about cases where Claude has proofread or edited human text?）— Anthropic 公式発表より

"A translation produced by Claude carries a watermark, because in this case every word is chosen by Claude."（Do watermarks apply to translations?）／"Detecting a watermark also doesn't work well on small samples, where there are fewer word choices and thus less information to go on. As a passage increases in length, confidence about Claude's involvement increases too."（Which specific method of watermarking do you use?）／"Light editing probably won't remove the watermark completely; a complete rewrite where every word is replaced will. In the latter case, of course, it's arguable whether the text can any longer be described as AI-generated."（Can't someone just edit the text to get around the watermarking?）— Anthropic 公式発表より

"Using our key, one can only answer the question “What is the likelihood this was partly written by Claude?” It doesn't confirm whether the text was human-written, and it can't tell whether the text was written by a different AI (even if that other AI uses watermarking, it would have a different key; it might also use a different watermarking method altogether)."（Which specific method of watermarking do you use?）／"A watermark can only determine that Claude was likely involved with the content at some point."／"It cannot distinguish “Claude wrote this” from “Claude heavily edited this.”"（いずれも What does a watermark actually prove?）／"A watermark only helps test whether Claude might have produced or processed the content. It doesn't say anything about ownership or authorship, and doesn't change a user's rights under our terms."（Does this change who owns a given output…?）— Anthropic 公式発表より

"We are releasing a detection API in private preview. It is currently available to eligible organizations as required under EU law (such as regulators, law enforcement, media, fact-checkers, independent researchers, educational organizations, and EU civil society groups)."（How do I check if a piece of text was written by Claude?）— Anthropic 公式発表より

"Claude may not be the original author. People often use Claude to proofread, translate, summarize, or convert files."／"Lack of a detected mark doesn't mean the content wasn't AI-generated or processed."／"It was generated by a model released before marking was supported;"／"The text has been heavily edited, paraphrased, translated, or mixed into other writing;"／"The passage is very short, leaving too little text for a reliable signal;"／"A file's metadata was stripped through format conversion, re-saving, screenshots, or other means;"／"It was produced through a platform, feature, or file type where a particular marking type wasn't supported."（いずれも Limitations）— Anthropic サポート文書より

"AI detection software uses a different method, because the companies that provide it don't have our key. Among other things, those services look at aspects of the text like the subtle (and not-so-subtle) "tells" that often appear in AI's phrasing. For example, AI models appear to be fond of the construction "this isn't [X], it's [Y]", and use the word "quietly" a lot more than you might expect. Picking up on these patterns is fundamentally different from checking for a watermark."（How does this differ from AI detection software, like Pangram?）— Anthropic 公式発表より

"When Claude produces a file of a supported type (such as a .png, .jpg, or .svg), it will attach a content credential in the form of a small, cryptographically signed note in the file's metadata, saying that the file was made or processed with Claude."／"This is an open industry standard called C2PA"／"Any C2PA-aware tool can read it; we'll be providing our own where you can drop a file and check."／"This metadata label is very different from a watermark. Nothing in the file changes—it is not embedded or hidden. As with text, the credential only says Claude was involved in producing the file; it doesn't include any identifying information."（いずれも What about images and other files?）— Anthropic 公式発表より

"This metadata follows the Coalition for Content Provenance and Authenticity (C2PA) open standard, which is used across the industry to record information about content provenance. If a signed metadata label is present, it signals that a file was processed by Claude and lets you detect whether the file has been tampered with."（2. Signed provenance metadata）— Anthropic サポート文書より

"Anthropic, along with several other major AI model providers and around 190 total signatories, signed the EU Code of Practice on Transparency of AI-Generated Content in July 2026. This requires AI system providers to use methods of "marking" AI-generated text."（Why are you watermarking Claude's outputs?）／"Watermarking won't be specific to Claude. As of August 2, the EU requires AI providers serving its market to mark AI-generated content. Other major model developers have signed the same Code of Practice and will be implementing their own watermarks."（To summarize）— Anthropic 公式発表より

"Providers of AI systems, including general-purpose AI systems, generating synthetic audio, image, video or text content, shall ensure that the outputs of the AI system"（Article 50(2)）— EU AI法の条文より

"One of my fundamental problems with this is that no two synonyms carry the exact same meaning."／"The idea that anything other than my needs should factor into the generation of text for me is patently offensive."／"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 presents to its users to read."（いずれも Objections to the Technical Premise 節）— John Gruber, Daring Fireball（2026年8月16日）より

"When a supported Claude model generates text, it weaves an imperceptible watermark directly into the text itself. You won't see it, and it doesn't change the meaning, quality, or readability of Claude's response."（1. Embedded watermarks in text 節）— Anthropic サポート文書より

"Yet the same provision exempts systems performing an "assistive function for standard editing", or which do not substantially alter the user's input or its meaning."／"Anthropic has nevertheless chosen a blanket, model-level implementation that appears broader than the law's minimum requirement. That may be convenient compliance engineering, but it discards distinctions the law expressly attempted to preserve. The result is a signal broad enough to implicate harmless and assistive use, yet fragile enough to be removed by a motivated person through substantial recomposition. It risks concentrating suspicion on ordinary and assistive users while remaining weakest against deliberate deception."／"But a watermark that burdens the candid and yields to the deceptive is not meaningful transparency. It is compliance theatre."— James Padolsey（2026年8月12日）より