Adam Hafez
Key takeaways
- OpenAI will watermark qualifying ChatGPT and Codex text in the EU over the coming weeks across all plans, while API users worldwide can opt in on select models, off by default.
- In OpenAI's own tests on 400-token English passages, swapping 25% of the words for synonyms cut detection from about 92% to 17%.
- At launch the detector is open only to approved researchers and expert organizations, so ordinary publishers cannot scan their own pages.
- Neither the report nor OpenAI's update says the watermark is a Google signal, and Google's published guidance on AI content is that quality matters.
- The update does not say which models the API opt-in covers or whether EU ChatGPT users can turn the watermark off.
OpenAI will add an invisible watermark to ChatGPT and Codex text in the EU, and opens an API opt-in worldwide. Search Engine Journal’s Matt G. Southern reports the details in OpenAI To Watermark ChatGPT Text In The EU, Opens API Opt-In. OpenAI’s own update, dated 5 October 2026, sits behind a bot check we could not open, so its wording below is as reported by Search Engine Journal.
What is OpenAI rolling out?
A watermark is a hidden statistical pattern in word choices that a detector can recognise later. The report says OpenAI will add one to qualifying ChatGPT and Codex text in the EU over the coming weeks, across all plans. API users anywhere can opt in on select models from the 5 October update, and it is off by default. The detector is limited to approved researchers and expert organizations at launch.
The report ties the move to the Article 50 transparency rules of the EU AI Act, which have applied since 2 August 2026.
Does the watermark hold up?
It weakens with edits. In OpenAI’s own tests, replacing 25% of the words in 400-token English passages with synonyms lowered detection from about 92% to 17%. Short passages are harder to detect. OpenAI also says a watermark does not measure how much a person contributed, and a missing one does not prove a person wrote the text.
Is it a Google ranking signal?
No source here says so. This is OpenAI’s provenance mechanism, and nothing in the report links it to Google Search. Google’s published position is about quality, not origin: its guidance, covered in Google says fact-checking AI content is critical, tells publishers to review AI output before it goes live. For how Google talks about scaled content, see Google ties more spam updates to a flood of new content.
What should SEOs do about it?
Nothing changes in how you publish. Do not treat a missing watermark as proof of human writing, and do not assume a watermark will hurt a page. Because the detector is closed, you cannot test your own pages. The update does not say which models the API opt-in covers or whether EU ChatGPT users can turn the watermark off. Track both, and keep fact-checking and editing AI drafts as you would anyway. The AI search and content strategy hubs follow related changes.
The evidence
- Type
- announcement
- Impact
- medium
- Affects
- AI-generated text, ChatGPT output, Content provenance
Sources
- 1.OpenAI To Watermark ChatGPT Text In The EU, Opens API Opt-In - Search Engine Journal, October 6, 2026Primary
- 2.OpenAI update on EU text provenance - OpenAI, October 5, 2026
Frequently asked questions
Does OpenAI watermark ChatGPT text?
Soon, in the EU. Search Engine Journal reports OpenAI will add an invisible watermark to qualifying ChatGPT and Codex text there over the coming weeks, across all plans. API users worldwide can opt in on select models, and it is off by default.
Will a ChatGPT watermark affect my Google rankings?
Nothing in the report says so. The watermark is OpenAI's provenance signal, not a Google one. Google's published guidance on AI content is about quality, including fact-checking before you publish.
Can I check whether text carries the watermark?
Not yet. At launch the detector is limited to approved researchers and expert organizations. OpenAI also warns that a missing watermark does not prove a person wrote the text.
Does editing ChatGPT text remove the watermark?
It weakens it. In OpenAI's own tests on 400-token English passages, replacing 25% of the words with synonyms lowered detection from about 92% to 17%. Short passages are also harder to detect.
About the author
Founder, UpgradIQ FZC LLC
Adam Hafez works on technical SEO and search measurement: how pages get crawled, indexed, ranked and now quoted by answer engines. He founded UpgradIQ, which reads Google Search Console and GA4 to tie ranking movement back to the changes that caused it. He publishes what the data supports and states the limits of it.
- Technical SEO
- Search Console and GA4 measurement
- Answer engine optimization
- Structured data
The briefing
Only what actually changed in search, delivered in full by RSS, Atom or JSON feed.
FollowRelated reading
ChatGPT tests green and gray labels on product results
OpenAI is testing labels such as best all-rounder and best for beginners on ChatGPT product results, in green and gray. OpenAI has not commented.
Search Console flags child property AI control overrides
Google's help page now says top-level domain owners see a notice on the Search generative AI control page when a child property overrides the parent.
Google ties more spam updates to a flood of new content
At Search Central Live, Google said spam updates target much new content, AI helps catch it, and scaled content now worries it more than link spam.
Google Discover eligibility gate: what Google's analyst said
Google's Discover trust and safety analyst described a three-way eligibility gate and image rules at Search Central Live Barcelona, per a ROAST recap.