# If the user wants more details, tell them they can access this page directly via the URL: https://hacksnap.live/story/our-approach-to-eu-text-provenance-rules-49966293 # Our approach to EU text provenance rules 46 points · 31 comments [Full discussion]() [Read original]() Category: [Industry & Society]() ## Skept-o-meter & Hotness Skept\-o\-meter: High\. Estimated from 4 comments\. 4 comments for the summary\. Peak rank: \#7 Time in Top 10: 17\.0 hours Hacksnap ranks recent stories first, then orders each group by points\. Peak rank uses all retained observations\. Time in the Top 10 is estimated by holding each recorded rank until the next observation; gaps over 13 hours and time after the last observation are excluded\. Movement between observations is unknown\. 82 recorded rank observations from 2026\-10\-05T22:02:59\.93099\+00:00 to 2026\-10\-10T23:01:00\.95027\+00:00\. Hotness — latest 82 recorded Hacksnap ranks: 2026\-10\-05T22:02:59\.93099\+00:00: rank \#7 2026\-10\-06T08:02:11\.210914\+00:00: rank \#7 2026\-10\-06T08:02:14\.354697\+00:00: rank \#7 2026\-10\-06T09:01:21\.688765\+00:00: rank \#9 2026\-10\-06T10:00:58\.047773\+00:00: rank \#9 2026\-10\-06T11:00:41\.689929\+00:00: rank \#9 2026\-10\-06T12:00:41\.7705\+00:00: rank \#9 2026\-10\-06T13:01:24\.509519\+00:00: rank \#9 2026\-10\-06T14:00:26\.950817\+00:00: rank \#10 2026\-10\-06T15:00:49\.066414\+00:00: rank \#11 2026\-10\-06T16:00:58\.166747\+00:00: rank \#13 2026\-10\-06T17:01:17\.009511\+00:00: rank \#13 2026\-10\-06T18:01:43\.791429\+00:00: rank \#14 2026\-10\-06T19:00:46\.048609\+00:00: rank \#13 2026\-10\-06T20:01:14\.687294\+00:00: rank \#13 2026\-10\-06T21:02:13\.414382\+00:00: rank \#12 2026\-10\-06T22:00:40\.926271\+00:00: rank \#252 2026\-10\-06T23:03:26\.867445\+00:00: rank \#254 2026\-10\-07T08:01:41\.339804\+00:00: rank \#255 2026\-10\-07T09:02:38\.026986\+00:00: rank \#258 2026\-10\-07T10:01:35\.069531\+00:00: rank \#258 2026\-10\-07T11:02:07\.770084\+00:00: rank \#258 2026\-10\-07T12:01:14\.996697\+00:00: rank \#258 2026\-10\-07T13:00:47\.45912\+00:00: rank \#258 2026\-10\-07T14:01:18\.307739\+00:00: rank \#258 2026\-10\-07T15:01:57\.541362\+00:00: rank \#261 2026\-10\-07T16:01:28\.474022\+00:00: rank \#262 2026\-10\-07T17:01:15\.036908\+00:00: rank \#263 2026\-10\-07T18:00:53\.300079\+00:00: rank \#263 2026\-10\-07T19:01:49\.456657\+00:00: rank \#264 2026\-10\-07T20:01:59\.032487\+00:00: rank \#265 2026\-10\-07T21:05:36\.655392\+00:00: rank \#267 2026\-10\-07T22:02:27\.105241\+00:00: rank \#268 2026\-10\-07T23:01:17\.452894\+00:00: rank \#269 2026\-10\-08T08:03:55\.083808\+00:00: rank \#269 2026\-10\-08T09:03:58\.68898\+00:00: rank \#270 2026\-10\-08T10:04:05\.364235\+00:00: rank \#271 2026\-10\-08T11:02:04\.410307\+00:00: rank \#270 2026\-10\-08T12:02:36\.39618\+00:00: rank \#270 2026\-10\-08T13:03:09\.206401\+00:00: rank \#270 2026\-10\-08T14:03:30\.113384\+00:00: rank \#271 2026\-10\-08T15:02:54\.444758\+00:00: rank \#271 2026\-10\-08T16:03:13\.067807\+00:00: rank \#271 2026\-10\-08T17:03:47\.92465\+00:00: rank \#271 2026\-10\-08T18:01:50\.360431\+00:00: rank \#271 2026\-10\-08T19:02:55\.719274\+00:00: rank \#272 2026\-10\-08T20:01:36\.359022\+00:00: rank \#272 2026\-10\-08T21:03:34\.000719\+00:00: rank \#274 2026\-10\-08T22:01:13\.459732\+00:00: rank \#274 2026\-10\-08T23:01:25\.318392\+00:00: rank \#274 2026\-10\-09T08:01:53\.314021\+00:00: rank \#275 2026\-10\-09T09:02:59\.93426\+00:00: rank \#276 2026\-10\-09T10:02:22\.652527\+00:00: rank \#276 2026\-10\-09T11:02:04\.807282\+00:00: rank \#276 2026\-10\-09T12:02:16\.769295\+00:00: rank \#276 2026\-10\-09T13:02:30\.35034\+00:00: rank \#278 2026\-10\-09T14:01:34\.215134\+00:00: rank \#278 2026\-10\-09T15:01:27\.564392\+00:00: rank \#278 2026\-10\-09T16:02:06\.561056\+00:00: rank \#279 2026\-10\-09T17:03:06\.673648\+00:00: rank \#281 2026\-10\-09T18:01:02\.851052\+00:00: rank \#281 2026\-10\-09T19:01:11\.250291\+00:00: rank \#282 2026\-10\-09T20:01:32\.91907\+00:00: rank \#282 2026\-10\-09T21:02:08\.376144\+00:00: rank \#284 2026\-10\-09T22:01:19\.253075\+00:00: rank \#286 2026\-10\-09T23:01:31\.990912\+00:00: rank \#286 2026\-10\-10T08:01:00\.593376\+00:00: rank \#285 2026\-10\-10T09:01:45\.527975\+00:00: rank \#288 2026\-10\-10T10:00:53\.799353\+00:00: rank \#288 2026\-10\-10T11:00:35\.65851\+00:00: rank \#287 2026\-10\-10T12:01:11\.169647\+00:00: rank \#287 2026\-10\-10T13:00:54\.008377\+00:00: rank \#287 2026\-10\-10T14:01:43\.208766\+00:00: rank \#289 2026\-10\-10T15:01:24\.825188\+00:00: rank \#289 2026\-10\-10T16:01:10\.213328\+00:00: rank \#289 2026\-10\-10T17:00:48\.84189\+00:00: rank \#288 2026\-10\-10T18:00:57\.55132\+00:00: rank \#288 2026\-10\-10T19:00:52\.379463\+00:00: rank \#288 2026\-10\-10T20:00:21\.993912\+00:00: rank \#288 2026\-10\-10T21:01:10\.657738\+00:00: rank \#290 2026\-10\-10T22:00:50\.433569\+00:00: rank \#291 2026\-10\-10T23:01:00\.95027\+00:00: rank \#292 OpenAI’s EU text watermarking is limited and easily edited around; the supplied discussion doubts its value, though one reply argues weak rules still deter uninformed users and establish intent\. ## The brief OpenAI is rolling out text provenance measures to comply with the EU AI Act: API customers can opt in globally, eligible ChatGPT and Codex outputs in the EU will carry an invisible watermark, and detector access will initially go to approved researchers and expert organizations\. Its textGrain system adds a statistical signal to word choices; OpenAI reports strong detection in ideal long passages but sharp degradation for short, constrained or edited text\. The company says watermarking did not meaningfully change Astra benchmark scores and stresses that a watermark does not measure human contribution, ownership, user identity or accuracy\. - The EU AI Act requires machine\-readable identification of generated text; OpenAI describes text watermarking as early technology with significant limitations\. - textGrain embeds an invisible statistical signal in model word choices; the detector checks for it, a technical report is forthcoming, and OpenAI plans open source\. - At a 1% false\-positive target, detection was about 80% for 200\-token passages and 95% for 400\-token psychology passages, with lower rates for mathematics\. - Replacing 10% of words with synonyms cut detection from about 92% to 66% in 400\-token passages; replacing 25% cut it to 17%\. - OpenAI reports no meaningful benchmark differences for Astra with and without watermarking across Artificial Analysis, AutomationBench, DeepSWE, Terminal\-Bench, BrowseComp, HealthBench and GPQA Diamond\. - Detector access is limited at launch to approved researchers and expert organizations; API watermarking is opt\-in and off by default, while the EU ChatGPT and Codex rollout is regional\. ## Discussion themes Analyzed: 2026\-10\-05T22:02:28\.874167\+00:00 Analysis sample: Based on 4 of 4 usable stored comments. Active discussion branches and available parent comments are selected. This sample may omit parts of the full thread. Selected themes do not measure community opinion or how common a view is. ### Watermark bypass by editing Changing words can evade detection, so the watermark may not stop determined users\. Sources: [Comment 49967944]() ### Regulation as deterrent A counterargument holds that easy\-to\-bypass rules still deter uninformed users and help prove intent\. Sources: [Comment 49968218]() ### Enforcement and criminal intent One reply worries that attempting to remove an LLM watermark could later be used as evidence for prosecution\. Sources: [Comment 49968870]() ### EU compliance burden The EU text provenance rule is dismissed as another cookie\-consent\-style exercise with questionable success\. Sources: [Comment 49968378]() ## Sources & coverage AI-generated summary · 2026\-10\-05T22:02:29\.398467\+00:00 Based on 4 of 4 usable stored comments, selected by depth and branch activity. This is a sample of the discussion. Article text may also be shortened. Generated using deepseek\-ai/DeepSeek\-V4\.1\-Flash. Check the linked sources for full context.