OpenAI rolls out invisible text watermarking for ChatGPT and Codex in EU to comply with AI Act
OpenAI will begin adding invisible watermarks to text outputs from ChatGPT and Codex in the European Union over the coming weeks to comply with the EU AI Act. The watermark, called textGrain, embeds statistical signals in word choices and is detectable only by OpenAI's proprietary tool, currently restricted to approved researchers. Detection rates drop significantly with editing—swapping 25% of words reduces detection to 17%. API customers worldwide can enable watermarking today for select models.
Editorial responsibility
- No named human review is recorded for this page.
- Reports are grouped by semantic similarity and deterministic rules. Language models may assist titles, summaries, translation and cross-source analysis; the page reads the event directly, while its address stays stable when the title changes.
- Summary covers the current reports
Cross-source coverage
Reporting timeline
OpenAI to watermark ChatGPT and Codex text in EU to comply with AI Act
OpenAI has announced it will watermark text generated by its AI models ChatGPT and Codex in the European Union to comply with the bloc's AI Act. The company noted that editing the watermarked text can make the invisible marks harder to detect. The move is part of broader efforts by AI developers to meet transparency requirements under the EU's forthcoming regulatory framework for artificial intelligence. The watermarking is intended to help identify AI-generated content, though OpenAI acknowledges that determined users may be able to circumvent the measure through editing.
Read sourceOpenAI to invisibly watermark ChatGPT and Codex text in the EU to comply with AI Act
OpenAI announced it will invisibly watermark text generated by ChatGPT and Codex for users in the European Union, in response to Article 50 of the EU AI Act. The watermark is embedded in the pattern of word choices, not special characters, and is invisible to readers. It can only be detected by OpenAI's proprietary detector, which is currently kept internal and shared only with approved researchers and expert organizations. The system, called textGrain, tilts the model's word choices to leave a statistical signal. In tests, the detector caught about 80% of 200-token psychology passages and 95% of 400-token ones at a 1% false positive rate. However, detection drops significantly with editing: swapping 10% of words for synonyms cut detection from 92% to 66%, and 25% swapping cut it to 17%. Mathematics answers scored lower due to tight wording. The watermark does not identify users, prompts, or owners. Eligible users on all plans will receive the watermark over the coming weeks, while API customers worldwide can enable it today for select models. The public cannot use OpenAI's detection tool.
Read sourceOpenAI to add invisible watermarks to ChatGPT and Codex outputs in the EU
OpenAI has announced that it will begin adding invisible watermarks to text outputs from its ChatGPT and Codex models in the European Union over the coming weeks. This policy action is aimed at enhancing traceability and potentially addressing concerns about AI-generated content. The announcement was made via a post on the Polymarket platform, indicating a proactive step by the company to comply with or preempt regulatory expectations in the EU. The specific technical details of the watermarking method and its impact on users have not been disclosed in this initial statement.
Read sourceShow 6 older updatesHide older updates
OpenAI admits text watermark detection drops to 17% after 25% word rewrite
A post on X by user DanDr1s reports that OpenAI has admitted its text watermark detection system is ineffective against simple rewrites. According to the post, swapping 25% of the words in a watermarked text causes the detection rate to plummet to 17%. Additionally, access to the detector is initially restricted to approved researchers only. The author speculates that the tool was developed primarily to comply with the EU AI Act rather than to function as a robust detection mechanism. The post links to an external source for further details.
Read sourceOpenAI to watermark ChatGPT and Codex outputs in EU to comply with AI Act
OpenAI will begin adding invisible text watermarks to outputs from ChatGPT and Codex in the European Union over the coming weeks, in response to the EU AI Act. The watermark is embedded in the wording itself, and OpenAI claims tests show no meaningful quality loss. However, the author expresses disapproval of detection requirements influencing how a writing tool phrases text. The post notes limitations: in one test on 400-token passages, replacing 25% of words with synonyms reduced watermark detection from about 92% to 17%. Additionally, the watermark cannot distinguish how much a human contributed; even text written by a human and edited by ChatGPT can carry it. The author sarcastically criticizes the EU for focusing on this issue.
Read sourceOpenAI expands content provenance to text with watermarking for EU AI Act compliance
OpenAI announced it is expanding its content provenance approach to include text, responding to EU regulatory requirements under the EU AI Act. The company acknowledged significant limitations of current text watermarking technology. OpenAI already provides tools to verify whether an image or audio file was created with its models, and this new effort builds on those capabilities to help people better understand when content may have been generated or edited with an OpenAI model. In the European Union, OpenAI will start watermarking eligible text from ChatGPT and Codex over the coming weeks to comply with the EU AI Act. Additionally, customers using OpenAI's API can enable text watermarking for select models worldwide starting today.
Read sourceOpenAI to roll out text watermarking for ChatGPT and Codex in the EU in coming weeks
OpenAI announced it will roll out text watermarking for its AI models ChatGPT and Codex in the European Union in the coming weeks. The watermarking feature will remain off by default in the API. The company stated that across benchmarks used to assess its latest frontier model, Astra, it does not see meaningful performance differences with and without watermarking. The announcement was made via an X post, which also included a comment from the poster suggesting the feature needs real testing.
Read sourceOpenAI adds invisible watermarks to ChatGPT and Codex text output in the EU
OpenAI announced it is adding invisible watermarks to text output from eligible ChatGPT and Codex models in the European Union. The watermarking feature will remain off by default in the API. Additionally, OpenAI is opening access to its text watermark detector for approved researchers and expert organizations. The move aims to make AI-generated text traceable, though the post frames it as a question to the audience. The announcement covers a specific policy change for EU users and a broader access initiative for the detection tool.
Read sourceOpenAI Unveils textGrain Watermarking for EU AI Act Compliance, Rolls Out to API and ChatGPT
OpenAI has announced textGrain, a text watermarking technology designed to comply with the EU AI Act. The system embeds invisible statistical signals into the word selection process of AI models. Starting immediately, API customers can selectively enable watermarks on certain models. Within the next few weeks, OpenAI will add invisible watermarks to outputs from ChatGPT and Codex for users in the European Union. The accompanying detection tool is currently only available to approved researchers and expert institutions. OpenAI also released detection performance data and limitations, allowing readers to assess the current capabilities and constraints of the text watermarking approach.