{"rewrite":{"id":"r_5d692c1ea0da235eb0b0362c","clusterId":"c_afda7af1d26ffc9d2b92b1e2","slug":"openai-to-watermark-chatgpt-text-in-the-eu","model":"deepseek-v4-1-flash","headline":"OpenAI to Watermark ChatGPT Text in the EU","summary":"OpenAI will add invisible watermarks to text generated by ChatGPT and Codex to comply with the EU AI Act. The feature rolls out within weeks for eligible plans in the EU, and API developers can enable it on some models, though it is off by default. Watermarking works by subtly varying word choices. OpenAI's detection rate drops from 92 percent to 66 percent when 10 percent of words are swapped for synonyms, and to 17 percent at 25 percent.","whyItMatters":"OpenAI is shipping a compliance feature while its own detection numbers show the watermark survives copy-paste but not light editing, which puts the burden of proof back on the EU's labeling rule rather than on the tool.","webCardHtml":"\u003cp\u003eThe watermark lives in word choice, not metadata. OpenAI says the pattern is invisible to readers and persists when text is copied and pasted, because it is built into the words themselves. The company also published a report on textGrain, a technology for determining whether that watermark is present.\u003c/p\u003e\u003cp\u003eDetection is limited. OpenAI reports that swapping synonyms erodes it, and that short sentences, math answers, and translated text are also hard to detect. Access to the detector will start with approved researchers and specialized institutions. OpenAI warns that a missing watermark does not prove human authorship, and a present one cannot measure how much editing or judgment went into the text.\u003c/p\u003e","blueskyPost":"OpenAI's own numbers put ChatGPT watermark detection at 66 percent once just 10 percent of words are swapped, and 17 percent at 25 percent. The EU AI Act compliance framing outruns what the method holds up to.","twitterPost":"OpenAI's watermark survives a 10 percent synonym swap at 66 percent detection. At 25 percent it falls to 17 percent.","threadsPost":"OpenAI's detection rate drops to 66 percent when 10 percent of words are swapped for synonyms, and 17 percent at 25 percent. That number sits inside the same rollout meant to satisfy the EU AI Act, and it says the watermark tracks light edits, not rewrites.","newsletterBlurb":"OpenAI will begin watermarking ChatGPT and Codex text in the EU within weeks, a requirement of the AI Act. The mark is built into word choices and survives copy-paste, but OpenAI's own report shows detection falls from 92 percent to 17 percent once a quarter of a sentence is reworded. Detector access starts with approved researchers and institutions.","attributionJson":"[{\"source\":\"GIGAZINE\",\"url\":\"https://gigazine.net/news/20261006-openai-start-watermarking-chatgpt-text-eu/\",\"title\":\"OpenAI Begins Efforts to Watermark ChatGPT Text\"}]","lintFlagsJson":null,"lintHits":0,"costUsd":0,"inputTokens":4629,"outputTokens":613,"status":"published","repairAttempts":0,"nextRepairAt":null,"factsAttemptedAt":1791270430,"createdAt":"2026-10-06T07:04:15.000Z","publishedAt":"2026-10-06T07:06:08.000Z","updatedAt":"2026-10-06T07:06:08.000Z"},"cluster":{"id":"c_afda7af1d26ffc9d2b92b1e2","canonicalTitle":"OpenAIがChatGPTのテキストに透かしを入れる取り組みを開始","representativeArticleId":"a_07f58935c8a8bb607b58a962","sourceCount":1,"writtenSourceCount":1,"writeAttempts":0,"isSolo":true,"entitiesJson":"{\"anime_titles\":[],\"manga_titles\":[],\"work_titles\":[],\"studios\":[],\"people\":[],\"type\":\"news\",\"domain\":\"other\",\"is_roundup\":false}","contentType":"news","status":"published","firstSeenAt":"2026-10-06T01:54:00.000Z","lastSeenAt":"2026-10-06T01:54:00.000Z","updatedAt":"2026-10-06T07:06:06.000Z"},"attribution":[{"source":"GIGAZINE","url":"https://gigazine.net/news/20261006-openai-start-watermarking-chatgpt-text-eu/","title":"OpenAIがChatGPTのテキストに透かしを入れる取り組みを開始"}],"entities":{"anime_titles":[],"manga_titles":[],"work_titles":[],"studios":[],"people":[],"type":"news","domain":"other","is_roundup":false},"keyFacts":null}